Upload
others
View
9
Download
0
Embed Size (px)
Citation preview
Epenthesis and deletion
LING 451/551
Spring 2011
Epenthesis and Deletion
• schema for epenthesis rules
0 X / (in some context)
• schema for deletion rules
X 0 / (in some context)
Iraqi Arabic
„my‟
binit binti „daughter‟
riʤil riʤli „leg‟
ʔisim ʔismi „name‟
ʧiðib ʧiðbi „lie‟
siʕir siʕri „price‟
ʔuxut ʔuxti „sister‟
ʃuɣul ʃuɣli „work‟
xubuz xubzi „bread‟
ʕumur ʕumri „age‟
kuʃuk kuʃki „cabin‟
Morphological analysis
„my‟
binit bint-i „daughter‟
riʤil riʤl-i „leg‟
xubuz xubz-i „bread‟
ʔuxut ʔuxt-i „sister‟
Iraqi Arabic words: root(-suffix)
Alternations
• Alternating forms of morphemes
– [binit] ~ [bint] „daughter‟
– [riʤil] ~ [riʤl] „leg‟
– [xubuz] ~ [xubz] „bread‟
– [ʔuxut] ~ [ʔuxt] „sister‟
– etc.
• Alternating segments
– i~0
– u~0
Distribution of alternants
CVCVC form of root
___#
CVCC form of root
___V
Possible analyses
• Deletion analysis
– URs • /binit/ „daughter‟, /binit-i/ „my daughter‟
– Deletion: high vowels 0 / VC ___ CV
• Epenthesis analysis
– URs • /bint/ „daughter‟, /bint-i/ „my daughter‟
• /xubz/ „bread‟, /xubz-i/ „my bread‟
– Epenthesis: 0 i / iC___C#
u / uC___C#
Deciding between epenthesis and
deletion
• Either rule describes the data provided
• Neither rule is especially complex or
implausible
• More data
„my‟
ʔibil ʔibili „camels‟
kutub kutubi „books‟
Morphological analysis
„my‟
ʔib-i-l ʔib-i-l-i „camels‟
kut-u-b kut-u-b-i „books‟
-i „my‟
-i/u- plural
Iraqi Arabic words: ro(-infix)ot-suffix
Prediction of deletion analysis
(output of morphology)
UR /ʔib-i-l-i/ „my camels‟
Deletion 0
PR *[ʔibli]
Epenthesis required for alternating forms
Root URs
bint „daughter‟
riʤl „leg‟
ʔism „name‟
ʧiðb „lie‟
siʕr „price‟
ʔuxt „sister‟
ʃuɣl „work‟
xubz „bread‟
ʕumr „age‟
kuʃk „cabin‟
ʔibl „camel‟
kutb „book‟
Distinctive features for vowels
i u
ɑ (not seen in data) (with [back] instead of [front]) i u ɑ [high] + + - [back] - + + [round] - + - [low] - - + many possibilities: [high], [back]; [high], [round]; [back],
[round]; [back], [low]; [round], [low]
Affix URs
• -/i/ „my‟:
V
+high
-back
• -V- plural
+high
Epenthesis, using features
0 i / iC___C# i.e. 0 V / V C___C#
[+high]
[-back] [-back]
u / uC___C# i.e. 0 V / V C___C#
[+high]
[+back] [+back]
“collapsible” (statable as a single rule)
0 V / V C___C#
[+high]
[back] [back]
Iraqi Arabic summary
• Morphemes
– Bound
• -i „my‟
• -i/u- pl.
– Roots
• /CVCC/
• High vowel epenthesis
More X~0 alternations
• Problems involving epenthesis and
deletion can be tricky
– element of uncertainty in morphological
analysis when X~0 at morpheme boundary
Cree
sg/prox plural obviative
[si:si:p] [si:si:pak] [si:si:pa] 'duck'
[mi:kis] [mi:kisak] [mi:kisa] 'bead'
[astis] [astisak] 'mitten'
[mo:s] [mo:swak] [mo:swa] 'moose'
[mostos] [mostoswak] 'buffalo'
[asa:m] [asa:mak] 'snowshoe'
[atim] [atimwak] [atimwa] 'dog‟
[pi:sim] [pi:simwak] 'sun'
[ayo:skan] [ayo:skanak] 'raspberry'
[amisk] [amiskwak] [amiskwa] 'beaver'
Data on downloadable handout
Morphological analysis
singular plural obviative
[si:si:p] [si:si:p-ak] [si:si:p-a]
The plural suffix is at least -/ak/, and the obviative
suffix is at least -/a/.
What about „moose‟?
[mo:s] [mo:s-wak] ?
[mo:sw-ak] ?
[mo:s-w-ak] ?
[mo:s-wa] ?
[mo:sw-a] ?
[mo:s-w-a] ?
All we can really tell:
„moose‟ is at least [mo:s]
plural and obviative suffixes are at least -/ak/ and -/a/,
respectively
Morphological status of [w] (~0)
Could be:
a) part of suffix morphemes (i.e. -/wak/, -/wa/)
deleted in certain contexts
b) part of (certain) root(s), e.g. /mo:sw/-,
deleted in certain contexts
c) not part of either morpheme, e.g. /mo:s-ak/,
epenthesized in certain contexts
What are the alternating morphemes?
Morphological analysis (a): [w] part of suffix
-a ~ -wa
-ak ~ -wak
Morphological analysis (b): [w] part of root
mo:s ~ mo:sw etc.
Morphological analysis (c): [w] not part of any morpheme
no alternating morphemes
Distribution of alternants
Morphological analysis (a) ([w] part of
suffix)
Alternating morphemes: -[ak] ~ -[wak], -[a]
~ -[wa]
All roots are non-alternating
/si:si:p/ /si:si:p-wak/ /mo:s/ /mo:s-wak/
W-deletion 0
[si:si:p] [si:si:pak] [mo:s] [mo:swak]
W-deletion: w 0 / (context to be determined) [w] 0
k___a p___a
s___a s___a
m___a m___a
n___a
No natural classes
Look farther for conditioning?
[w] 0
sk___a i:p___a
os___a is___a
im___a a:m___a
an___a
Still no natural classes
Nothing that seems likely to condition [w] deletion
Now what?
Distribution of alternants
Morphological analysis (b) ([w] part of
(certain) root(s))
Alternating morphemes
some roots: [mo:sw] ~ [mo:s]
Non-alternating morphemes
other roots: [si:si:p]
all suffixes: -[ak], -[a]
/si:si:p/ /si:si:p-ak/ /mo:sw/ /mo:sw-ak/
W-deletion 0
[si:si:p] [si:si:pak] [mo:s] [mo:swak]
W-deletion: w 0 / (context to be determined)
[w] 0
___V ___#
/w/ 0 / ___ #
Distribution of alternants
Morphological analysis (c) ([w] not part of
any morpheme)
No alternating morphemes
roots: [mo:s], [si:si:p]
suffixes: -[ak], -[a]
/si:si:p/ /si:si:p-ak/ /mo:s/ /mo:s-ak/
w-epenthesis w
[si:si:p] [si:si:pak] [mo:s] [mo:swak]
0 [w] / (context to be determined) [w] 0
k___a p___a
s___a s___a
m___a m___a
n___a
Deciding between analyses
• Morphological analyses (a) ([w] part of suffix), (c) ([w] not part of any morpheme) – Distribution of [w] cannot be predicted
[w] 0
k___a p___a
s___a s___a
m___a m___a
n___a
• Morphological analysis (b) ([w] part of certain roots) – Distribution of [w] is phonologically predictable
– /w/ 0 / C ___ #
Summary of analysis
/w/ Deletion:
w --> 0 / ___ # (probably C ___#)
URs:
• Affixes:
-/ak/ plural
-/a/ obviative
Root URs • /Cw/-final roots:
/mo:sw/ „moose‟
/mostosw/ „buffalo‟
/atimw/ „dog‟
/pi:simw/ „sun‟
/amiskw/ „beaver‟
• /C/-final roots:
/si:si:p/ „duck‟
/mi:kis/ „bead‟
/astis/ „mitten‟
/asa:m/ „snowshoe‟
/ayo:skan/ „raspberry‟
In Cree, there is an underlying contrast between roots which end in
/Cw/ and those which end in a single consonant. The contrast is
neutralized word-finally in favor of the single consonant.