Melody

1
Melody

• Perfecto Herrera
• Music Perception and Cognition

2
Representation of Melodic Knowledge

• Psychophysical function (frequencies -gt pitches)
• Pitches (all pitches usable in a musical culture)
• Tuning system (available pitches for building
  melodies)
• Modal Scale (some pitches are assigned roles,
  weighted with different importance, etc.)

3
Functional units

• Melodic grouping several pitched events that are
  perceived as a unit (similarity of interval and
  direction)
• Stream coherent grouping of pitched events that
  separates from other groupings because of their
  pitch distance and/or speed of events
• Phrase group of melodic groupings that can be
  fit within STM (or can be played in a breathing
  cycle) (usually bounded by silences) a
  continuous gesture-
• Contour shape or outline of a melody
• Melodic schema an archetypal (or prototypical)
  melodic contour that can be elaborated and
  ornamented to form many melodies
• Tonality sensation that there is a pitch that
  acts as a center or magnet of the pitches used in
  the melody

4
Melodic grouping

• Horizontal (pitch space) and longitudinal (time)
  process of assigning events to a given stream
  (mostly bottom-up)
• Vertical similarity plus temporal proximity
  influences grouping
• Other influences
• Glides (connectedness favors grouping)
• Progressions (elements changing in a particular
  direction tend to be grouped)
• Timing (regular recurrence favors grouping)

5
Melodic motion

• Size and direction of pitches determine melodic
  motion
• Size steps (small size) versus leaps (big size)
• Direction up, down, no-change
• Two pitches create a context were a third one can
  be more or less expected
• Continuity (similar intervals in the same
  direction) versus Reversal (one or both aspects
  changes)

6
The Implication-Realization Model

• Proposed by Narmour (1990, 1992), and partially
  supported by research by Krumhansl (1991, 1995),
  Schellenberg (1996, 1997), Cuddy Lunney (1995)
  results not affected by musical training,
  familiarity, or musical culture (Chinese, Sami
  cultures also tested)
• The model is based on Gestalt processes involved
  in closure of melodic phrases. Closure
  interacts with grouping/phrasing. Narmour claims
  that the same principles can be applied to
  duration, timbre and intensity.
• The Implication Realization Model suggests 5
  cues that determine closure
• A temporal gap or rest in the music
• A move from a less stable to a more stable pitch
• A move from shorter to longer note durations
• A metrical emphasis
• Melodic Reversal the core of the IR theory

7
The Implication-Realization Model

• Key elements Register (Contour) and Interval
• Register is either Up, Down or Lateral i.e. U, D,
  L
• Interval is Unison upwards i.e. U, m2, M2, m3,
  M3,.P8, m9, M9,.
• These are formulated in two principles
• Registral Continuation / Reversal
• Intervallic Similarity /Dissimilarity
• These provide the basis for describing Process
  and Reversals
• Process means Stability (X-X -gt X)
• Reversal implies Closure (X-Y -gt Z)
• Take two successive pitches P1 and P2, this
  interval implies a third one, that can be
  realized as the implication indicated or can be
  a negation of it
• Examples
• If the interval between them is lt to 5 semitones
  
• Then the implication consists of
• The Registral Direction (up/down) will remain the
  same
• The Interval will remain similar, i.e. lt m3
• If the interval between them is gt 7 semitones
• Then the implication consists of
• The Registral Direction will change
• The Interval will become dissimilar, i.e. lt M2

8
The Implication-Realization Model

• However, the implication is only prospective
• There is also what actually happens and this is
  only judged retrospectively
• So, after another note has occurred the
  implication is either
• Realized
• Process P (continuation)
• Reversal R (change)
• Partially denied
• Intervallic Duplication ID
• Intervallic Process IP
• Registral Process VP
• Intervallic Reversal IR
• Registral Reversal - VR
• Completely Denied
• (P) (R) (ID) (IP) (VP) (IR) (VR)

9
Melodic implication
Reversal of direction and interval size
Continuity of interval size and direction
Continuity of direction Reversal of interval size
Continuity of interval size Reversal of direction
10
Melodic implication

• IP intervallic process
• ID intervallic duplication
• VP registral process
• IR intervallic reversal
• VR registral reversal

• D duplication
• P process
• R reversal

11
Melody Schemata, Attention and Memory

• Music understanding is a dialogue between
• Experience and expectation - derived from memory
• Structure of what is being perceived - what is
  occurring now
• Experiments with interleaved melodies show that
  we can hear more when we know what to expect and
  listen for (top-down influence) and also that
  like other gestalts we can hear one or the other
  but not both. This is consistent with streaming
  principles.
• Expectations are also referred to as schemata

12
Schemata

• Schemata are archetypal organisations or
  conventions that have been abstracted from
  salient and/or statistically significant aspects
  in the structure of objects or events
• We compare these schemata with the present and
  they determine to a large extent what we can and
  cannot expect

13
Music schemata
3 different melodic schemata, based on Meyer
(1958)

• Dynamic knowledge structures
• Some of them related to physical or visual
  concepts
• Help to recognize and to code a series of events
  or objects
• Help to predict possible next musical notes,
  directions of contour, etc
• Tuning systems and scales do not include temporal
  information
• Tonality is one of the most powerful music
  schemas as it includes, implicitly, temporal
  dependencies derived from the sequential pattern
  of notes

14
Melodic processing

• Overall - 4 aspects of pitch used in melodic
  processing
• Pitch Context
• Contour
• Key Distance
• Intervals and Chroma

15
Pitch Context in Melodic Organisation

• Krumhansl (1979) found that remembering a pitch
  is affected by the tonality of its context.
• Played a standard tone, a pause, then an
  interfering pitch sequence, another pause then
  another tone, i.e. Tone, pause, Sequence, pause,
  Tone
• The distracters were either atonal or in the
  scale of C. The pitch either belonged to the
  scale of C or not. The second tone was either the
  original tone or a semitone up or down. Was the
  2nd tone the same as the first?
• E.g. G C D E A F G C F E A F
• G C D E A A G C F E A F
• Results
• Therefore if there is a conflict between a
  stimulus and the expectations associated with the
  schema, performance worsens. The schema activated
  by the distracters interferes in memory.

16
Pitch Context in Melodic Organisation

• Guildford Hilton (1933) if a melody is played
  and then a note is changed, not only is it
  perceived to shift but its neighbours also shift.
• Dewan, Cuddy Mewhort (1977) played tonal and
  atonal 7 note sequences followed by a pair of
  probe tones. One was present in the sequence one
  was not
• Question which tone was in the sequence much
  better performance for tonal sequences
• Cuddy Cohen and Miller (1979) 3 note sequences,
  which were changed. Changes better recognised
  when fragments was in a larger consistent tonal
  context
• Conclusion Scales are schemas that scaffold
  musical memory

17
Melodic processing

• Overall - 4 aspects of pitch used in melodic
  processing
• Pitch Context
• Contour
• Key Distance
• Intervals and Chroma

18
Contour

• Contour is the rise and fall between pitches
  (e.g. 0 1 2 -3 or --UP-UP-DOWN)
• Experiments indicate that changing the contour
  affects recognition
• Melodies that share contour but differ in
  intervals or pitches are harder to differentiate
  than those with different contours.
• This is particularly so for atonal melodies where
  there is no tonal (pitch set) schema being used
  (the most memorable characteristic for these type
  of melodies
• Frances (1958) recognition of transpositions
  from same contour imitations better from tonal
  rather than atonal melodies (i.e. coherent from
  incoherent pitch set)
• Dowling and Fuujitani (1971) compared melodies
  transformations (listen X, then listen Y and Z
  and decide which one is identical to X)
• Straight transpositions
• Same Contour but changed intervals
• Different Contour
• a) and b) were distinguished from c) 85-90 of
  the time
• As distinguished from Bs only 50 - i.e. at
  chance levels
• Dowling 1978 Confirmed Dowling Fujitani 1971
  and found that trained musicians are better with
  atonal variation -- presumably have more
  developed pitch schemata

19
Intervals Pitch Chroma in Melodic Organisation

• Patterns of interval are however important and
  seem to be more so when the tune is well known
• Recognition of same contour imitations is at
  90 for well known tunes but 70 for unknown
• It seems that coding of melodies varies depending
  on whether Short Term or Working/Long-Term Memory
  is involved
• LTM needs more precision to be able to make
  distinctions between lots of similar items so
  it uses chroma and intervals STM can cope quite
  well exploiting just the contour

20
Intervals Pitch Chroma in Melodic Organisation

• Dowling Bartlett (1981) compared STM LTM,
  using pairs of melodies A A, B B
• (A B B A)
• A B were either
• transposition (all intervals preserved)
• tonal imitation (same contour, different
  intervals)
• changed contour
• Subjects had to judge transformation B to be
  type a), b) or c), then to make same judgement
  for A
• Assumption is that comparing A to A requires LTM
  (rather than STM, which was involved in the B to
  B answer)

21
Intervals Pitch Chroma in Melodic Organisation

• Results
• Inner pairs transpositions imitation were
  distinguished from changed contour (75 72 )
  but not from each other
• Outer Pairs performance worse also more
  difference between transpositions imitation
  (65 57) inner to outer pairs success
  dropped by 10 for transpositions, 15 for
  imitations
• This could imply that
• chroma interval are coded in LTM before contour
  (if contour was also coded then the drop between
  inner and outer pairs would be the same)
• interval/chroma LTM coding requires more time,
  rest, sleep, etc.

22
Key Distance in Melodic Organisation

• If the tonality plays a role in recognising
  differences and similarity of melodies then the
  distance between keys should be important (This
  is predominantly an issue for Western tonal
  music)
• Keys share more or less notes between 1/7 to
  6/7
• Bartlett and Dowling (1980) constructed pairs of
  melodies that were more or less closely related
  by key
• Tonal sequences vs.
• 1. near-key imitations
• 2. far-key imitations
• 3. atonal imitations
• 4. different contour
• far-key as easy to reject as atonal and different
  contour imitations
• Transposition to far keys share fewer pitches --
  makes rejection of distortions easier this is a
  negative property it did not help recognition
  of transpositions to far keys.
• Key distance effects confirm importance of scale
  schemata
• Contour and pitch set means that we dont have to
  remember exact intervals

23
Coding of melodies

• Contour interval information two extraction
  processes, variable, context-dependent robustness
  (one related to global, holistic processing
  right hemisphere- the other to local,
  language-related processing left hemisphere)
• Scale could provide a framework to make their
  combination more effective than when separated
• Again partial redundancy, mutually supportive
  parallel processes (as in pitch perception)

24
Recognition of familiar melodies

• Recognizing a well-known melody (e.g., one's
  national anthem) is not an all-or-none process
• Instead, recognition develops progressively while
  the melody unfolds over time
• The gating paradigm present segments of
  increasing duration (i.e., the first note, then
  the first two notes, then the first three notes,
  and so forth). Recognition was assessed after
  each segment either by requiring participants to
  provide a familiarity judgment or by asking them
  to sing the melody that they thought had been
  presented.
• The more familiar the melody, the fewer the notes
  required for recognition.
• Musicians judged music's familiarity within fewer
  notes than did nonmusicians, whereas the reverse
  situation (i.e., musicians were slower than
  nonmusicians) occurred when a sung response was
  requested
• Both musicians and nonmusicians appeared to
  segment melodies into the same perceptual units
  (i.e., motives) in order to access the correct
  representation in memory.
• Cohort model (Marslen-Wilson, 1987) applied to
  the music domain (i.e. progressive accumulation
  of evidence to discriminate among melodic
  candidates)

25
The Development of Melodic Organisation

• Pitch seems to be first coded in an absolute way,
  exposure helps to develop relative pitch coding
  (depending on the native tongue!)
• Contour schemas appear first, followed by tonal
  schemas
• Infants (five months old!) can recognise melodic
  contour (measured by heart rate changes to
  transpositions versus new tunes)
• Chang Trehub (1977) - Experimented with 5
  month old babies 6 note atonal melodies
  repeated 30 times baby adapts and ignores new
  melody played either a transposition or a
  permutation of the original. So one preserved
  contour the other didnt.
• Babies reacted to changed contour - i.e.
  permutation
• At this age intonation (prosodic) contours of
  language are thought to be important. Different
  types of messages could share prototypical
  melodic contours (request, attention grabbed,
  denial)

26
The Development of Melodic Organisation

• 3 months - babies have octave equivalence
• 6 months - babies are quite accurate at imitating
  the pitch of isolated notes
• Up to 2 years pitch matching, tune recognition
• singing simple phrases
• invention and spontaneous songs
• regular rhythmic beating patterns
• Acquisition of skill varies widely across
  individuals and sibling imitation and education
  are important 2nd and subsequent children get
  taught by older siblings

27
The Development of Melodic Organisation

• 2 years More singing,
• general exploration of vocal capacities.
• Single phrases repeated over and over
• Discrete steps between focal points not tonal
• Contour is predominant melodies shifted around
  but contour preserved
• Development of songs seems to go along with
  simple stories
• 5-6 years Sense of key starts to emerge
• Imperfect recognition of pitch shifts
• Principle of conservation applied to melodies
  less sensitivty to surface changes
• 6-7 years Imberty (1969) much better with pitch
  shifts
• By 7 changes in key are recognised in tunes
• 8 years Difference between major and Minor
  scales is recognised