minimal sonority distance requirementscourses.washington.edu/lingclas/451/msd_skt_gk.pdf · 2011....
TRANSCRIPT
Minimal Sonority Distance
requirements
LING 451/551
Winter 2011
Sanskrit vs. Ancient Greek clusters
• Word initial, medial onset clusters differ
– Medial onset clusters not driven by
Maximal Onset Principle
• Sonority Hierarchy for both lgs. – glide - liquid – nasal – fricative - stop
• Minimum sonority distance requirement
Ancient Greek (400 B.C.)
• 2 types of “onset” clusters
– “Core” onset clusters
• word-internal or word-initial
– “Marginal” onset clusters
• word-initial only
Core onset clusters
r l n
p pn
t tl
k kl kn
g gr
Core onset clusters in Greek
Onset Formation
σ
|
Cx Cy
where Cx is less sonorous than Cy
(Also Minimum Sonority Distance requirement,
below)
Marginal onsets
m s p t d
s sm sp
z zd
p ps
k kt
Note: Sonority decrease: [sp], [zd]
Sonority plateaus: [kt]
Marginal onset clusters
• created by special adjunction
word[ σ
|
C C
no sonority requirement
Why 2 types of onset clusters
• 1. Perfect formation
– /Ce/- (C reduplication) with core onset
clusters
• /kleph/ ke-klopha 'steal'
– /e/- alone with vowel initial roots
• /angel/ e-angelka 'announce' [a:ngelka]
– /e/- with marginal onset clusters
• /kten/ e-ktona 'kill'
Analysis of perfect formation
• Form core syllables
• (klo)(pha)
• k(to)(na)
• Required of core onsets – Sonority increase
– Minimal sonority distance of two
glid liq nas fric stop
glid
liq
nas
fric
stop kl kt
Perfect formation
• Prefix /e/- to all forms; syllabify – e-klopha: (e)(klo)(pha)
– e-ktona: (ek)(to)(na)
• Reduplicate stem initial consonant if syllable initial – (e)(klo)(pha) (ke)(klo)(pha)
– (ek)(to)(na
• Unusual case of P (syllabify) before M (reduplication) – our model: M > P
2. Stress differentiates core,
marginal onsets
• For medial core onset
– V1 in V1CCV2 scans as “light”: i.e. V1.CCV
(e.g. V.grV) (open syllable)
• For medial CC with same segments as
word-initial marginal onset
– V1 in V1CCV2 scans as “heavy”: i.e. V1C.CV2
(e.g. Vs.mV) (closed syllable, must be
accented)
Sanskrit onsets
• Like Greek
– Core vs. marginal onset clusters
• Different from Greek
– Different MSD requirement
Sanskrit core onsets • Core onset clusters need only be separated by
one step of Sonority Hierarchy – glide - liquid – nasal – fricative - stop
• Triconsonantal: k
r m ñ s ʂ s sm
ś śr
p ps
d dr
ʤ ʤñ
k kʂ
Sanskrit marginal onsets
• Fricative+stop [st, śc]
– only word-initially, by special adjunction
Summary of Greek, Sanskrit
• Word-internal syllabification obeys Son
Hier, Minimum Sonority Distance
Requirement
– 2 steps: Greek
– 1 step: Sanskrit
• Word edge syllabification
– Special adjunction (not subject to Son Hier,
MSD requirements)
Ancient Greek
• Data on class web site
• Mycenean (Linear B)
– samples:
http://www.explorecrete.com/archaeology/line
arB.pdf
"Determine what decision the [orthographic]
system makes to deal with the problem posed by
consonant clusters."
• More generally, what is the mapping between
phonology, orthography?
• Word-final consonants are not written
– knossos <ko-no-so>
– deksiwos <de-ki-si-wo>
– ksenwos <ke-se-nu-wo>
• The first consonant in certain consonant sequences is not written – aleksandra <a-re-ka-sa-da-ra>
– phasgana <pa-ka-na>
– phulakphi <pu-ra-pi>
– knossos <ko-no-so>
• The consonant sequences which are not represented orthographically consist of – nasal+stop
– s+stop
– stop+stop
– s+s
clusters with no sonority increase from C1 to C2
• The high vowel [i] is not written in:
– aiksma <a-ka-sa-ma>
• If [i] in [aiksma] is really a glide [j], then
another C not written is
– j + stop
– another cluster with no sonority increase from
C1 to C2
• The first consonant in other clusters is written (with copy of following vowel) – aleksandra <a-re-ka-sa-da-ra>
– knossos <ko-no-so>
– deksiwos <de-ki-si-wo>
– ksenwos <ke-se-nu-wo>
– aiksma <a-ka-sa-ma>
– dosmiya <do-si-mi-ya>
• These consonant sequences consist of: – stop + s
– stop + nasal
– stop + s + nasal
– stop + liquid
– s + nasal
– nasal + glide
clusters in which sonority increases from C1 to C2
"Determine what decision the [orthographic]
system makes to deal with the problem
posed by consonant clusters."
• In the first type of cluster, not represented
orthographically, there is a syllable
boundary between the consonants,
whereas in the second type of cluster, the
consonants are tautosyllabic.
• Onsets, but not codas, are
orthographically represented.
"Does this result shed any light on the syllabic
structure operative in this dialect of Greek?"
• Apparently words were syllabified – knos.sos de.ksi.wos
– kse.nwos a.le.ksan.dra
– phas.ga.na phu.lak.phi
– ai.ksma do.smi.ya
• Permitted onset clusters consisted of
glide liquid nasal fric stop
glide
liquid
nasal nw
fric sm
stop dr kn ks
• Assuming Son Hier – glide - liquid – nasal – fricative - stop
• Then MSD requirement = 1 in Mycenean
– 1 step separation: [sm], [ks]
– 2 step separation: [nw], [kn]
– 3 step separation: [dr]
“Assuming that the Attic dialect discussed in section 6.5 is
a later descendant of Mycenean, can you say anything
about whether Sanskrit or Attic syllabification more closely
reflects the system that must have operated in PIE?"
Attic (4th c. BC) Mycenean
(1000-2000 BC)
Sanskrit
nw
śr
ks, ksm kʂ, kʂɳ, ps
kl, tl
gr dr dr
sm sm
kn, pn kn ʤñ
Attic (4th c. BC) Mycenean
(1000-2000 BC)
Sanskrit
MSD = 2 MSD = 1 MSD = 1
• Simplest historical scenario
PIE (MSD=1)
Mycenean (MSD=1) Sanskrit (MSD=1)
|
Attic (MSD=2)
• Attic has innovated by imposing a more
stringent requirement on clusters
• Alternative scenario
PIE (MSD=2)
Mycenean (MSD=1) Sanskrit (MSD=1)
|
Attic (MSD=2)
• 3 historical changes required