Language and Music
in Optimality Theory
Outline
• Structural resemblance between language
and music
• Claim: every form of temporally ordered
behaviour is structured the same way
• Claim: insights of music theory can help out
in phonological issues
• Rate adjustments in language and music:
compression or restructuring?
Jackendoff and Lerdahl
• Jackendoff & Lerdahl (1980) point out the
resemblance between the ways both
linguists and musicologists structure their
research objects.
• Lerdahl & Jackendoff (1983) A Generative
Theory of Tonal Music, MIT Press,
Cambridge, Massachusetts.
Synthesis of linguistic methodology and the
insights of music theory
A Generative Theory of Tonal Music
• Description of how a listener (mostly
unconciously) constructs connections in the
perceived sounds
• The listener is capable of recognizing the construction of a
piece of music by considering some notes/chords as more
prominent than others
cf. Language
• Our cognition thus works in a way
comparable to how a reader divides a text
(often unconciously too) into different parts
A Generative Theory of
Tonal Music
(Lerdahl & Jackendoff, 1983)
• The research object is structured
hierarchically and in each domain the
important (heads) and less important
(dependents) constituents are defined by
preference rules
• Preference rules determine which outputs,
i.e. the possible interpretations of a musical
piece, are well-formed
Preference Rules
• Preference rules indicate the optimal
interpretation of a piece. Some outputs
are more preferred than others
• Preference rules, however, are not strict
claims on outputs. It is even possible for a
preferred interpretation of a musical piece
to violate a certain preference rule as long
as this violation leads to the satisfaction of
a more important preference rule
cf. Optimality Theory
(Prince & Smolensky 1993)
• Optimality Theory is a theory of
language and grammar in which wellformedness constraints on outputs
determine grammaticality.
• These constraints apply simultaneously
to representations of structures. They
are potentially conflicting and they are
soft, which means violable.
Structuring of the Domains
Tuxedo Junction
motif
phrase
section
Prosodic Construction of a Phrase
x
x
x
x
x
x
x x x x x x
Mis sis sip pi Del ta
s w s w s w
w
s
s
w
syllable level
foot level
phrase level
Comparison structuring rules
• Music:
The domains in the music theory are called Timespans: Rhythmical units constructed from the
interaction of the metrical structure and the
grouping structure.
• metrical structure
(lower domains)
:
=
=
...
• melodic/harmonic (or grouping) structure
(higher domains):
motif < phrase < section ...XTC:
English
roundabout
Comparison structuring rules
• Language:
phoneme <  (syllable) <  (foot) <  (phrase)
p
pa
papa
de oude papa
phoneme < morpheme < word < compound
p
-pje
bloempje
muurbloempje
Comparison Preference Rules
Comparison preference rules 1
• Music (time-span reduction preference rule 1):
Choose as the head of a time-span the chord
(or the note) which is in a relative strong
metrical position (= the first position in a
measure)
• Language:
Choose the first  in a  as the head
Arguments for trochaic feet
Neologisms:
Acquisition data:
Cito, Prolog, Brinta
Mispronunciations:
narcis, parfum
1;6
Comparison preference rules 2
• Music (time-span reduction preference rule 2):
Choose as the head of a time-span the chord
(or the note) which is relatively harmonically
consonant (segmental markedness)
• Language (peak prominence):
Choose as the head the heaviest available
syllable
Comparison preference rules
• Language:
Peak Prominence: stress the heaviest
available syllable: CVVC; CVCC > CVC; CVV
ki.dhar
as.baab
reez.ga.rii
> CV
sa.mi.ti
ru.kaa.yaa
aas.maan.jaah
Stress assignment in Hindi: Peak Prom. >> Nonfinality
Comparison preference rules
• Music (time-span reduction preference rule 2):
Choose as the head of a time-span the chord
(or the note) which is relatively harmonically
consonant (segmental markedness)
C
> C7
> … > Csus4
> Cdim
Over smaak valt te twisten
C vs C0
•
C
>
Cdim
Fifth C - G
0.5
0
-0.5
0
0.02
Time (s)
0
C vs C0
-0.5
0
•
0.02
C
>
Cdim
Time (s)
C - Gb
0.5
0
-0.5
0
0.02
Time (s)
C vs C0
•
C
>
Cdim
Fifth C - G
0.5
0
-0.5
0
0.02
Time (s)
C vs C0
•
C
>
Cdim
Fifth C - G
0.5
0
-0.5
0
0.02
Time (s)
0
C vs C0
-0.5
0
•
0.02
C
>
Cdim
Time (s)
C - Gb
0.5
0
-0.5
0
0.02
Time (s)
C vs C0
•
wave C+G
0.953
0
-0.953
0
0.0951548
Time (s)
C vs C0
•
wave C+Gb
0.9987
0
-0.9987
0
0.0945913
Time (s)
Comparison preference rules
• Music (time-span reduction preference rule 7):
Choose as the head of a time-span the chord
(or the note) which emphasizes the end of a
group as a cadence
tonic > dominant > subdominant > parallel ...
• cf. Language: Phrasal rule
C7-B
C7-F
cadence
Tonic - Dominant - Subdominant
• Examples of 3 chord songs:
mccoys - hang on sloopy
(russell & farrell)
royal guardsmen - snoopy vs. the red baron (gernhard & holler)
rolling stones - get off of my cloud
(jagger & richard)
grease soundtrack -summer nights
(jacobs & casey)
any trouble - second choice
(gregson)
sonics - psycho
(roslie)
standells - sometimes good guys don’t wear white (cobb)
r.e.m.- stand!
(buck, stipe, mills, berry)
rare breed - beg, borrow and steal
(difrancesco & zerato)
kingsmen - louie louie
(r.berry)
Time-span reduction
Mozart: Sonata K.331, I
Time-spans
Conflict TSRPR1 - TSRPR7
Conflict
The A6-chord is in a metrically stronger
position, but E-chord is harmonically more
consonant
constraints  TSRPR 7
TSRPR 2
TSRPR 1
candidates 
 E
A6
*
*!
*
First Language Acquisition Data
segmental & positional markedness: same preference
syllabe
pre-m.
s
onset
rhyme
margin
nucleus
m.core
k
b
t
satellite
l
r
peak

o
u
satellite
coda
k
d
l
app.
Segmental markedness: /s/ > /x/
Positional markedness: /x/ > /s/
syllabe
(2;0)
pre-m.
s
onset
rhyme
margin
nucleus
m.core
x
satellite
peak
a
satellite
coda
p
app.
*Complex >> Pos. Markedness >> Segm. Markedness
Conclusion
• Structural resemblance between language and
music
(cf. also Lasher (1978), Gilbers (1984, 1987), Mallen (2000), Gilbers &
Schreuder 2002))
Every form of temporally ordered behaviour
is structured the same way
Descargar

No Slide Title