Human Language, Segmental distribution of speech sounds in

Report
Human Language, Segmental distribution of
speech sounds in prosodic domain: Evidence
from 20th century Modern Phonological
paradigms and theories
Dr.Hemanga Dutta
Assistant Professor, Department of Linguistics and Contemporary
English, School of Language Sciences
The English and Foreign languages University, (EFLU)
Hyderbad: 500007
Mail id: [email protected]
Language from various perspectives: History,
Philosophy, Biology and Sociology
• What is Language?
• Language , pre structural paradigm and
philological concerns
• Language and Structural principles of Saussure
and B.F. Skinner
• Language, Cognitivism, Chomsky and
generative paradigm
• Language, Communication and Functional
prerequisites to competence model
Phonetic underpinnings of Human
linguistic mechanism
• Phonetics: three main components – Articulatory,
Acoustic and Auditory
• Articulatory phonetic paradigms: Air stream
mechanism, Process of Phonation, Oro nasal
process
• Speech organs
• Manner of articulation: Plosive, Fricative,
Affricate, Laterals, Approximants, Nasals
• Place of articulation: Bilabial, Labio dental,
Dental, Alveolar, Post alveolar, Retroflex, Velar,
Uvular, Glottal etc.
Phonology and Generative paradigm
• Phonology as organization of sounds in to
patterns
• Phonology as Functional Phonetics
• Difference between Phonetics and Phonology
• Integration model between Phonetics and
Phonology
• Classical Phonemic model: Phoneme,
Allophones, Contrastive and Complementary
Distribution and Free variation
Phonemic analysis of Lambada
• Contrastive Distribution: Contrastive distribution
refers to a process in which two phones can occur in
identical environment and the substitution of one
sound with another creates a semantic difference.
• /p/, /b/
• [pɑːɳi] - water
• [bɑːɳi]
broom
• [pa:p]
sin
• [baːp]
father
• Here /p/ is substituted by the phoneme /b/.
Free variation
Two phonetically identical sounds are always in free
variation. Where the one occurs the other can occur
without changing the meaning.
• /dʒ/, /ð/
• [kaːgədʒ] -paper
• [kaːgəð]-paper
• Here /dʒ/ is substituted by the phoneme /d/ which
does not bring change in meaning.
• /k/, /θ/
• [tʃoːkraː] -boy
• [tʃoːθraː] -boy
Complementary distribution
• When two sounds occur in mutually exclusive
environment they are called allophones
according to the principle of complementary
distribution.
• The distribution of clear /l/ and dark / / in
English
kli:n ‘clean’ lait ‘light’ lit ‘little’
Distinctive feature theory and the
issue of Binarism
• Distinctive feature theory the essence of which can be traced back
to Panini’s Pratyaharas, was first formalised by the Prague
Structuralists although it attained its prominence in the hands of
Chomsky and Halle in 1968 with the publication of SPE.
• Motivations:
• All segments must be characterizeable in terms of unique
combination of features. They can show the segment inventories of
the language.
• They should be able to show the segmental contrast in the world
languages.
• The segments must be characterizable in phonetic terms. This has
led to the requirement of the Natural ness condition (Postal 1968),
according to which distinctive features must have a phonetic
underpinnings.
Major class features
• [+/- sonorant]: [+sonorant] segments are produced with a constriction in
the vocal tract which allows an air pressure behind it and in front of it to
be relatively equal, while it is not the case for [-sonorant] segments.
[+sonorant] segments include vowels, glides, liquids and nasals while [sonorant] are plosives, fricatives, affricates and laryngeal segments.
• [+/- consonantal] : [+consonantal] segments have a constriction
somewhere along the centre line in the vocal tract which is at least as
narrow as that required for a fricative. [-cons] lacks such constriction.
[+cons] segments are plosives, affricates, fricatives, liquids and nasals. [cons] sounds are vowels, glides.
• [+/-approximant]: [+approx] segments allow frictional escape of air
(Ladefoged 1971, Clements 1989). Vowels and non nasal sounds like
laterals and rhotics are [+approx].
• [+/- syllabic]: segments which can become nucleus of a syllable are
[+syllabic].
• Laryngeal features:
• Laryngeal features imply the glottal
properties of the segments.
• [+/-voice]
• [+/-spread glottis]: [+s.g] segments are
characterized by a vocal cord configuration
that produces audible glottal friction.
Aspirated segments are [+s.g].
• Manner features:
• Manner features are based on the type of constriction
or the manner of articulation.
• [+/-continuant] : continuant sounds are produced
when the primary constriction in the vocal tract does
not block the air flow. In the case of non continuant
sounds the air flow through the mouth is stopped.
Plosives, affricates, nasal s, laterals are [-cont] whereas
vowels and fricatives are [+cont].
• [+/-nasal]
• [+/-lateral]
Place features
• [LABIAL] : segments are articulated with the lips and in the case of vowels
with lips rounding. LABIALS may be specified for [+/round].
• [CORONAL]: These segments are articulated with a raised tip or blade of
the tongue. t, d, s etc are [+coronal] where as p, b, m etc are [-coronal].
[CORONAL] segments are further specifies for the features [+/- anterior]
and [+/-distributed]
• [+/-anterior] sounds are articulated in front of the palato alveolar region
of the mouth..
• [+/-distributed] sounds are those during the production of which the
articulatory constrictions extends for a considerable distance along the
direction of the air flow. Apical consonants are [-distributed] whereas
laminal consonants are [+distributed].
• [DORSAL] sounds are articulated with dorsum. Velars and uvulars are
[+dorsal].
• [Radical] also known as [Pharyngeal] articulated with root of the tongue
Features relating to the body of the tongue:
• [+/-high]: when the body of the tongue is raised
above the neutral position, the resultant sound is
high.
• [+/-low]: Low sounds are produced when the
body of the tongue is lowered below the neutral
position.
• [+/-back] : Retraction of the back of the tongue
from the neutral position during the production
of a sound makes it a back sound.
• [+/-tense]: [+tense] vowels are produced with
considerable muscular effort.
Acoustic feature
• [+ /-strident] : [+strident] sounds are marked
acoustically by greater noise than non strident
sounds. Fricatives are [+strident] sounds.
DF and some questions
• Do DFs characterizing a language from part of a universal
inventory?
• Are the correlates of DF primarily articulatory, acoustic or both?
• Should DFs be binary or not?
• DF is conceived as a universal inventory capable of describing the
phonology of any language. One standard argument in favour of
universalism is that recurrent and finite nature of dimensions that
phonologists refer to in the analysis of phonological systems,
whether in terms of rules, phonotactic statements or in system
inventories.
• Although DFs are more abstract than their implementations, they
are grounded in phonetics.
Assamese phonotactics and initial onset
consonant cluster: sonority, optimality and
binding principles
• Stop + Liquid
• /pr/
pran
‘life’
•
prem
‘love’
• /pl/
plawon ‘shower’
•
plawito ‘filled with water’
• /br/
brɒhmandɒ ‘universe’
• /bʰr/
bʰrɒm
‘illusion’
• /bl/
blauz
‘blouse’
• /tr/
troyudox ‘thirteenth’
•
tritiyo
‘third’
• /dr/
drɒbjɒ
‘substance’
•
dristi
‘vision’
•
/kr/
krɒm
‘serial’
•
krɒmannoye ‘gradually’
• /kl/
klantɒ
‘tired’
• /gr/
grismɒ
‘summer’
•
grɒhɒn
‘eclipse’
• /gl/
glani
‘repentance’
• /gʰr/
gʰran
‘smell’
• /tj/
tjag
‘sacrifice’
• /gj/
gjan
‘knowledge’
The unpermitted initial consonant cluster comprising of stop and liquid are */tl/ and */dl/ although they
conform to the principle of sonority hierarchy.
Nasal +Liquid
• /mr/
mrityu
‘death’
mrigo
‘deer’
• /ml/
mlan
‘pale’
• /nr/
nritjɔ
‘dance’
• /nj/
njai
‘justice’
• In Assamese phonotactics the alveolar and velar
nasal can not constitute the cluster with liquid in
initial onset position of a syllable. So the
prohibited patterns are */nl/, */ŋr/ and */ŋl/.
•
•
•
•
•
•
Fricative + Liquid
/sr/
srɒm
‘labour’
sristi
‘creation’
sriŋɒ
‘peak’
/hr/
hridɒi
‘heart’
In Assamese phonotactics there are only two fricatives
/s/ and /h/ which can function as the initial member of
a cluster comprising of fricative and liquid.
• The prohibited patterns are:
•
*/zr/, */zl/, */hl/
•
•
•
•
•
•
•
•
•
•
•
•
Fricative + Stop
/ st/
stɒmbʰɒ
‘pillar’
stɒn
‘breast’
stʊti
‘prayer’
/stʰ/
stʰan
‘place’
stʰanijɒ
‘local’
/sp/
spɒrxɒ
‘touch’
spriha
‘desire’
/spʰ/
spʰʊliŋɒ
‘ashes’
/sk/
skrin
‘screen’
/skʰ/
skʰɒlɒn
‘degradation’
This is the special property of the Assamese post alveolar fricative /s/ that can
function as the initial member of a consonant cluster followed by voiceless stops.
But it cannot make cluster with the voiced plosives of Assamese. So, the prohibited
shapes in Assamese consonant cluster phonotactics are: */sd/, */sb/, */sg/
•
•
•
•
•
Fricative + Nasal
/sm/
smɒxam
‘graveyard’
/sn/
snigdʰɒ
‘calm’
snan
‘bathe’
The post alveolar fricative /s/ can form consonant cluster
with bilabial and alveolar nasal as initial member of the
consonant cluster. The velar nasal /ŋ/ never becomes the
member of any consonant cluster. What is interesting to
note that other ficative sounds /z/ and /h/ are not allowed
in Assamese phonotactics to be the members of a
consonant cluster with nasals. Hence the prohibited
consonant clusters are : */sŋ/, */zm/, */zn/,*/zŋ/, */hm/,
*/hn/, */hŋ/
• Word initial clusters such as /pl, gl, pr, tr, dr,
gr/ are found but such as */lp, rt/, etc, where
the sonority relations are reversed, are not
present in the phonotactics of Assamese. So, it
can be argued that sonority constraints play
an important role in the patterning of
segmental distribution in the phonotactics of
a language.
• This hierarchy of onsets yield markedness constraints on onset
sonority, as proposed in de Lacy (2001)
• *ONS/L >> *ONS/N>>*ONS/OIn Assmese, the ban on liquids as the
first member in word initial cluster can be represented by using the
constraints in OT framework in the following way:
•
*ONS-L >> IDENT-I0 [approx] insures that liquid onsets will be
eliminated.
• But word initial syllables in Assamese are allowed to begin with
nasal consonants. It can be shown by the following ranking of the
OT constraints:
• IDENT-IO [approx] >>*ONS/N
• Moreover, what is seen from Assamese onset
consonant cluster is that the second member
of the cluster agrees in terms of feature
[voice] with the first member of the cluster. It
can be formulated in the following fashion in
the form of a constraint: Given a consonant
cluster C1C2, if C1 is voiced, then C2 must be.
• Word initial onset cluster in Assamese can be formed by combining either
stop + liquid or nasal+liquid. But liquid does not have the potential to be
the initial member of the onset cluster. As, for instance the following are
not the onset cluster in Assamese permitted by the phonotactic
constraints in Assamese: *rp, *rt, *lp, *rk, *rm, *lm
The asymmetry here is motivated by the Complexity Condition: nasals
containing three elements have priority over liquids, which contain two or
sometimes only one element. Thus if we argue in lines of Harris (1990) it is
to be noted that sonority is structurally encoded where a correlation can
be established between sonority and segmental complexity. As we have
seen in the above illustration of Assamese data plosives govern liquids on
the ground that plosives are more complex than liquids in terms of
internal components involved in representation. According to Harris
(1990) the more sonorant a segment, the less complex its representation.
But Rice (1992) has argued that greater sonority implies greater
complexity.
• Harris (1990) interpretation:
Coronal stop
Coronal nasal
x
x
Coronal lateral
x
?
Rᵒ
Rᵒ
Rᵒ
?ᵒ
?ᵒ
hᵒ
N+
•
H• Here, ? = occlusion ; Rᵒ = coronal; hᵒ= noise and H- = stiff vocal cords
Representation of coronal stop, nasal and lateral in the framework of
Harris(1990)
Rice (1992) interpretation:
•
ROOT
ROOT
•
SL
AF
SL
Place
•
Place
SV
ROOT
AF
SL
Place
AF
SV
Lateral
• Here, SL= Supralaryngeal ; AF= Air Flow; SV= Sonorant Voice
Representation of coronal stop, nasal and lateral in the framework of
Rice(1992)
sonority profile within an onset is met only if the
second consonant has more SV structure than the
first.
Stop
Lateral
ROOT
ROOT
SL
SL
SV
Margin hierarchy approach and
consonant cluster arrangment
•
Margin hierarchy of Prince and Smolensky (1993), gives preference
to segments of low sonority. This constraint is applicable to
singleton onsets or to the first member of an onset cluster and it is
known as M1 hierarchy. In the same way, the M2 hierarchy applies
both to the second member of an onset and a singleton coda. It
differs from M1 hierarchy in the sense that it gives preference to
consonants of high sonority.
M1 hierarchy (preference given to consonants of low sonority)
*M1/r >> *M1/ l >> *M1/ Nas >> *M1/obs
M2 hierarchy (preference given to consonants of high sonority)
*M2/ obs >> *M2/ Nasal >> *M2/l >> *M2/r
• The *ONSET/X subhierarchy assumed here is
shown below:
• *ONS/GLIDE>>*ONS/RHOTIC>>*ONS/LATERAL>>
*ONS/NASAL>>*ONS/VOICED
• OBST>>*ONS/VCLSOBST
• Smith (2003) claims what is appealing about the
*ONSET/X subhierarchy is that, as this sub
hierarchy is based on the sonority scale and the
perceptual preference for low sonority onsets, it
is functionally grounded.
• This preference for low sonority onsets is
assigned functional motivation too. The
auditory system is particularly sensitive to
rapid changes in spectral patterns (Stevens
1989; Ohala 1992; Delgutte 1997; Warner
1998). Delgutte (1997) claims that a low
sonority onset is preferred because it is more
distinct from the syllable nucleus than a high
sonority onset would be.
Assamese phonotactics, Binding principle and
OCP constraint:
• In Assamese phonotactics we have observed that some
clusters are not permitted to occur in word initial position.
The cluster types are */tl/, */dl/, */sl/, */zr/, */zl/, etc. A
bound consonant contains dependent structure. i.e.
identical Place structure to the consonant that binds it or
no Place structure.
Binding is not allowed within an onset (i.e. a consonant can
not be syllabified into an onset if it shares place with the
adjacent syllabified consonant.
A consonant must be bound for Place heterosyllabically (i.e.
a consonant may be syllabified into a rhyme if it is non
distinct in Place from the following onset). (parametric).
• Binding: /p/ and /l/ differ in place of
articulation, and thus /l/ is not bound.
Hence, the syllabification is well formed.
But the prohibited consonant cluster such as
*/dl/ can be accounted for in the light of
binding principles.
• s as Branching onset (Carlisle 1988, Major
1996,2001, Ohala 1999)
•
σ
•
•
ONS R
•
• X X X
•
• s
t o
• s as Complex segment (Selkirk 1982, Lamontagne
1993, Van de Weijer 1996)
•
σ
•
•
•
•
•
s
ONS
R
X
X
t
o
• s as Adjunct (Barlow 2001, Barlow and Dinnnsen
1998, Kaye 1989, Kenstowicz 1994)
•
σ
•
•
•
X
s
ONS
R
X
t
X
o
• s as Extrasyllabic (Appendix) (Goad and Rose
2004, Fikkert 1994, Levin 1985,
Giegerich 1992)
•
σ
•
•
ONS
R
•
•
X
X
•
• s
t
o
• In opposition to the branching onset approach
the other alternative approaches ranging from
complex segment approach to extrasyllabic
approach aim at eliminating the SSP violations
via the assignment of abstract
representations.
• Moreover, sonority sequencing constraints such as syllable
contact treats s-obstruent clusters differently from
obstruent- sonorant clusters. s-obstruent clusters have
falling sonority; hence epenthesis at the edge is possible
and preferred. Consider the Assamese example:
•
skul > is.kul
•
skrin >is.krin
• In the above example of loan word in Assamese, epenthesis
occurs at the edge. The crucial assumption here is that the
default site of epenthesis in loan words is at the edge.
• It is observed that edge epenthesis violates
NO CODA and ONSET whereas the
dispreferred internal epenthesis satisfies NO
CODA, ONSET and SYLLABLE CONTACT. In
order to justify this edge epenthesis Gouskova
(2001) talks about the constraint CONTIGUITY
• CONTIGUITY: elements adjacent in the input must be adjacent in
the output. This constraint ensures edge epenthesis when SYLLABLE
CONTACT is not at stake. SYLLABLE CONTACT implies sonority must
not rise across a syllable boundary (Davis 1998, Hooper 1976,
Murray and Vennemann 1983, Vennemannn 1988).
/skul/
 is.kul
si.kul
*COMPLEX
DEP
*
*
SYLL CONTACT
CONTIGUITY
*!
• In the above tableau SYLLABLE CONTACT is not
violated and hence, CONTIGUITY ensures that
edge epenthesis is optimal. So from this tableau
it is revealed that SYLLABLE CONTACT determines
the epenthesis site: at the edge for falling
sonority clusters and internal epenthesis for
rising sonority clusters. This outcome can be
expected as long as SYLLABLE CONTACT is ranked
above CONTIGUITY, although its ranking in
relation to DEP is not crucial. This pattern is
termed as the Emergence of the Unmarked
effect.
Gemination
• Gemination as a process triggered by liquids and semivowels:
• In this section I am proposing a problem that in Sanskrit gemination of
obstruents is triggered by the following liquids and glides such as j,r,l,ʋ
but not other segments such as nasals and obstruents. jətr̪ ə - jət.̪ tr̪ ə
(here)
•
tə̪ tr̪ ə - tə̪ t.̪ tr̪ ə
(there)
•
ʃukrə - ʃuk.krə (venus)
•
bʰəd̪rə - bʰəd̪.d̪rə (descent)
•
nepətʰ̪ jə - nepət.̪ tʰ̪ jə (back stage)
•
karjə - kar.rjə
(work)
•
pəd̪jə - pəd̪.d̪jə
(poetry)
•
pəkʋə - pək.kʋə
(ripe)
•
sətʋ̪ ə - sət.̪ tʋ̪ ə
(white)
•
tə̪ tʋ̪ ə - tə̪ t.̪ tʋ̪ ə
(theory)
•
ʃuklə - ʃuk.klə
(bright)
• The reason behind the motivation for the process of gemination triggered
by glides and liquids can be supported with cross linguistic observation.
For convenience, consider the internal representation of a word ‘sətjə’
(truth):
•
This internal representation can be made in two ways:
• If we consider word internal stop as a better candidate for the coda position
the representation will be as follows
•
•
σ
σ
•
•
•
s ə t̪ .
j
ə
•
Figure : Representation of word internal stop in coda position
• In contrast, stop and liquids form well formed consonant cluster cross
linguistically. From this perspective, the following will be the
representation:
•
•
σ
σ
•
•
•
s
ə . t̪ j
ə
•
• Figure No 1/E: Representation of word internal stop in the onset position
forming a cluster with following liquid
• From the above representations it is clear that either coda condition or well
formed onset cluster condition will be fulfilled. But from cross linguistic
evidence it is seen that alveolar and velar stops function not only as better
candidates for the coda position in a word but also form well formed
consonant cluster with liquids and glides.
• The fulfilment of these two conditions lies in the process of
gemination of the stops as shown in the following representation:
•
σ
•
•
•
•
•
•
s
ə t̪ . t̪ j ə
Figure No 1/F: Representation of word internal stop both in the
coda and the onset position
• But, what is interesting to note in this context that other sounds apart
from these liquids and semi vowels do not have the potential to
trigger gemination to the previous obstruent.
• As, for illustration consider the following instances:
• (5)
•
santi - *san.nti
•
ʃəbdə - * ʃəb.bdə
•
məndə - *mən.ndə
•
atma - *at.tma
• The motivation hidden behind the non occurrence of the process of
gemination of the obstruents followed by nasals and obstruents can be
assigned to the fact that obstruent and nasals do not serve as well formed
consonant cluster. Hence, it can be argued that glides and liquids have the
special property of being appropriate coda as well as a member in the
onset consonant cluster resulting in the process of gemination. In
addition, another generalization can be observed from the above process
of gemination that the asymmetry in phonological processes can be
attributed to the segmental properties and cross linguistic well
formedness conditions.
Conclusion
• segmental property inherent in a sound and
cross linguistic well formed conditions also play
a significant role in the triggering of certain
phonological processes.
• Behind the patterning of consonant cluster in a
specific manner the notion of phonological
strength and phonotactic patterns can be
realized.
THANK
YOU

similar documents