Lexical Phonology

Document Sample
Lexical Phonology Powered By Docstoc
					Lexical Phonology

Kiparsky: early 1980s

   Developing work by Dorothy Siegel,
    Steven Strauss, Mark Aronoff, David
A theory of many things...
A theory of the relationship between
  phonotactics and what once was called
  automatic morphophonology.
(But automatic morphophonology grew into
  all of what phonology was.)
A theory of levels or layers in morphology.
A constraint on neutralizations rules'
  application in nonderived environments
A theory of underspecification
   Lexical phonology is extremely
    derivational: things happen, some
    things happen before other things
    happen, and if something happens
    before X appears on the scene, then too
    bad for X. If Y isn't "in the lexicon", then
    a lexical rule can't "see" it (whatever
    that means).
   The most remarkable claim of lexical
    phonology is that the generalizations
    describing markedness principles of a
    language are the same as the rules
    governing the changes of sounds under
    conditions of word-formation.
Lexical/postlexical components

   This is the most important distinction,
    one going back a long way, but dropped
    for a while in generative phonology.
   Lexical rules have exceptions, do not
    create novel segments or sequences =
    morphophonemic rules.
   Postcyclic: flapping; word-final
    devoicing in German, Dutch, Russian;
    Cyclic = trisyllabic laxing.
Lexical/post-lexical split
   Lexicon                   No lexical
   Structure-preserving       exceptions
    (output is possible       not necessarily
    UR)                        structure-preserving
   Not necessarily           may apply across
    phonetically natural       words
   never applies across  May not refer to
    words                      internal
   Apply only in              morphological
    derived                    information
    environments              Flap formation
   Trisyllabic
           Ordered: Lexical >> Postlexical
Derived environments…the case
from Finnish
halut-a to want
halus-i  wanted
Non-derived environments:
tila     room
aiti     mother
Finnish (2): interaction with
ei/__ #
joki river     joke-na essive sg.
äiti mother äiti-nä     essive sg
vesi water     vete-nä essive sg
käsi hand      käte-nä essive sg
 Watch difference between äiti and vete.
Trisyllabic shortening (aka Trisyllabic
divine   divinity
serene serenity
profane profanity
vile     vilify
clear    clarity
rite     ritual
grade    gradual
tyrant   tyranny tyranize tyrannous
penal    penalize
fable    fabulous
Trisyllabic shortening
   nightingale, stevedore, ivory, Amory,
    bravery, mightily, pirating, obesity.
   So: 3SS has lexical exceptions, it
    doesn’t look over word-boundaries, and
    it seems to be a “markedness”
    statement for nonderived forms; it
    creates a segment type that exists

    V  [long] /__ CV [stress ]CV
Certain suffixes trigger the shortening: ity,
   ify, ual, ??ize (no), ??ous
Group 1: stress affecting:
-ic, -al, -ity, -ion, -y (nominalizing), -al, -
   ate, -ous, -ive, -ation
Group 2 stress neutral
-hood, -ful, -ly, -ize, -ness, -less, -y (adj.)
-able (Aronoff)
-ism (Goldsmith: philosophy 1, language-
Basis for stratal distinction:
   Group 1 easily attaches to non-word
    roots (e.g., paternal), while Group 2
    almost always attaches to existing
   Group 1, when it attaches to words, is
    stress-changing. Group 2 is stress-
    neutral, always…
   Group 1 make the resultant word look
    as much as possible like a (simple)
Combinations of Class 1,2

   Class 1 + Class 1: histor-ic-al, illumina-
    at-tion, indetermin-at-y;
   Class 1 + Class 2: frantern-al-ly,
   Class 2 + Class 2: weight-less-ness
   ?? Class 2 + Class 1: *weight-less-ity,
Derivational suffixes in English
able     fixable, doable,
ant      claimant, defendant
(at)ion  realization, assertion, protection
er       teacher, worker
ing      the shooting, the dancing
ing      the sleeping giant, a blazing fire
ive      assertive, impressive, restrictive
ment     adjournment, treatment,
ful      faithful, hopeful, dreadful
(i)al    presidential, national
(i)an Arabian, Einsteinian, Minnesotan
ic         cubic, optimistic, moronic,
ize        hospitalize, crystalize
ize        modernize, nationalize
less       penniless, brainless
ous        poisonous, lecherous
ate        activate, captivate
en         deaden, blacken, harden
ity        stupidity, priority
ly         quietly, slowly, carefully
ness       happiness, sadness
Strauss, Goldsmith ist-ic
 sexist     *sexistic     sadist     sadistic
 faddist    *faddistic    fascist    fascistic
 rightist   *rightistic   linguist   linguistic
 racist     *racistic     deist      deistic
 rapist     *rapistic     jurist     juristic
 putchist *putschist      sophist    sophistic
fatalist    fatalistic    masochist    masochistic
regalist    regalistic    plagiarist   plagiaristic
humorist    humoristic    populist     populistic
socialist   socialistic   atheist      atheistic
humanist    humanistic    aphorist     aphoristic
realist     realistic     hedonist     hedonistic
communist   communistic   anarchist    anarachistic
Word-final stress in base
   cartoonist   *cartoonistic
   escapist     *escapistic
   falangist    *falangistic
   alarmist     *alarmistic
   defeatist    *defeatistic
   adventist    *adventistic
   conformist   *conformistic
   extremist    *extremistic
   reservist    *reservistic
   careerist    *careeristic
Strauss notes:
-ic may not attach to an X+ist base if:
a. the final syllable of X is not primary
   stressed; and
b. X is a lexical item.
Stress clash in English

   OK morpheme internally:
   Revlon, nylong, Ticonderoga, Rangoon,
   Contrast alarmistic and admonish:
   admonish is like alarmist in stress, but
    admonition sounds fine, while
    alarmistic does not.
   abnormal -> abnormality
   2 1 3        2 2 133
Suffixes that care about stress:

   -ize:
   distinguish catholicize or notarize from
   Standardize and cannibalize suggest
    Stratum 2.
   Winterize, summerize, autumnize (cf.
    autumnal), but *fallize, *springize.
*New Yorkize
Perhaps what is not possible is:
[+stress] ]Stratum II [+stress]

that would suggest:
[ [ alarm ]stratum II ist ]stratum I ic
English stress, a bit
   Assuming the final syllable does not
    have a long vowel, then we can
    approximate English stress with the
    Latin main stress rule: stress on the
    penult if heavy, on the antepenult
   In fact, Group 1 causes a lot of
    changes, and Group 2 doesn’t, but we
    don’t have an explanation of that within
    lexical phonology
Problems, so far:
   In tough cases, we have to say that a
    suffix falls into both classes, which
    reduces the strength of the claim a lot.
    Typical cases are -able, -ism, -ize.
   Discussion of -ment in AMP.
   If -ize is Stratum 2 (always) then
    organization is a problem, since it
    violates the Level Ordering Hypothesis
    (Mark Aronoff’s observation). Perhaps
    likewise is ability, and istic.
GJ’s example of morphologically
sensitive rule (p. 120)
   Underlying           Surface        Gloss
   lo:p]verb + @n]inf   lo:p@          to walk

   za:k]noun + @n]Pl    za:ke          things

   [te:k @n] @n ] inf   te:k@ ne       to draw
   te:k@n]noun          te:k@          sign

   o:p@n]adj            o:p@           open

   te:k@n]verb          te:k@n         draw

   o:p@n]verb           o:p@n          open

 What is the alternative hypothesis?
Kenstowicz' Polish example

o >> u / __ +cons, -nasal, +voiced.

1. lexical exceptions: skrop
  "scratch"imper. from /skrob/
2. Morphological conditioning: more
  frequent in feminine and neuter nouns
  than in masculines:
   fem doz-a         pagod-a     mod-a
   with genitive plural: dus, pagut, mut;
   contrast with masc. mop, snop
   "Polish speakers are aware of the
    sound substitution effected by the
    raising rule...the voicing change is
    essentially below the level of
    consciousness." Careful!!
   Does not introduce any new phones
   "Raising of o to u before a voiced
    nonnasal consonant is an arbitrary and
    phonetically unmotivated sound
    substitution. " Careful!!