A computational analysis on the relationship between melodic originality and thematic fame in classical music from the Romantic period.
2025-04-30
0
0
1.01MB
7 页
10玖币
侵权投诉
A computational analysis on the relationship between melodic
originality and thematic fame in classical music from the
Romantic period.
Hudson Grith
grithh@u.edu
1 INTRODUCTION
Throughout modern society, many individuals have attempted to
predict popularity. Whether it be the popularity of a company, a
new product, or a tweet. This is often done by nding a correlation
between popularity and another variable. In some cases, it may be
analyzing media type to predict the popularity of an advertisement
[
21
] and in other cases, it is analyzing melodic originality to predict
the popularity of classical music [
13
]. This paper will be looking
into the relationship between melodic originality and popularity in
classical music. For this research, I would like to propose a novel
method for calculating melodic originality. This novel method will
be used in order to investigate the research question: To what extent
does melodic originality aect thematic fame in classical music from
the Romantic Period?
The research question proposed above will be answered through
a computer content analysis of 428 classical pieces from the Roman-
tic period. More specically, this analysis was performed in Python
[
11
] and R [
12
] by code I created specically for this analysis. For
this analysis, many of the denitions and methodology are based or
heavily derived from Dean Keith Simonton’s research on melodic
originality in the 1980s and 1990s. For this analysis melodic orig-
inality will be dened as the occurrence of uncommon two-note
transitions. This is very similar to Simonton’s denition which is
that melodic originality is “the occurrence both of chromatic notes
or less commonly used diatonic notes and of rare intervals between
consecutive notes” [
13
]. Despite the similar denition, this analy-
sis will be using a novel and dierent formula than Simonton so
it is expected for the results of this paper to dier on some level
from the conclusions stated in Simonton’s paper. This analysis will
use Simonton’s denition of thematic fame which he denes as
“the frequency that the theme would be heard in performance or
recording (i.e., the level of appreciation by concert goers, record
buyers, and performing musicians)”[
13
]. These two variables will
be compared to test their relationship and specically how well
melodic originality explains the variance of popularity. The novel
method for calculating melodic originality represented in this paper
addresses a gap in the research of the well-known researcher of
melodic originality, Dean Keith Simonton. To calculate originality
he only looked at the rst 6 notes in each piece. The benet of this
new method proposed in this paper is that it provides much more
information and data about each piece compared to Simonton’s
method; however, further research is required before the robustness
of this novel method is conrmed.
2 LITERATURE REVIEW
In this literature review key research in the eld of computational
musical analysis, melodic originality, and thematic fame will be
discussed.
2.1 Field Experts
A key expert in the eld of melodic originality in classical music
is Dean Keith Simonton, a distinguished professor of Psychology
at UC Davis. He denes much of the literature and methodology
to this day for analyzing melodic originality in a wide variety of
music. His 1980 paper “Thematic fame and melodic originality in
classical music: A multivariate computer-content analysis” was
the starting point for analyses into melodic originality of classical
music and the rst paper of its kind [13]. Simonton has since pub-
lished more papers relating to computational analysis of classical
music up to 1994 and his paper “Computer Content Analysis of
Melodic Structure: Classical Composers and Their Compositions”
[
15
]. His two research papers published in 1980 heavily focus on
melodic originality so my analysis will be more focused on these
papers. Research published in recent years continues to mention
Simonton and his formulas. An example of this is Richard Hass who
explored the relationship between melodic originality and fame in
20th-century American popular music [
3
]. In Hass’s paper, he used
an almost identical comparison for melodic originality to Simon-
ton, showing the wide impact of Simonton’s measure for melodic
originality; furthermore, according to Google Scholar his second
paper published in 1980 was cited by 207 other papers, once again
conrming the importance of his research. When it comes to the
broader eld of computational analyses of classical music there are
other researchers inuencing the eld. One being Christof Weiß
who works at the International Audio Laboratories Erlangen in Ger-
many. He has published several papers on computational analyses
in classical music, which unlike Simonton do not focus on melodic
originality or thematic fame. Weiß’s original doctoral dissertation
was on the topic of “Computational Methods for Tonality-Based
Style Analysis” [
19
] and he has continued research in this eld
specically looking at tonal complexity and tonality-based style
analysis [
18
]. He and his colleges make up a majority of recent
publications in classical music computational analyses [20].
2.2 Researcher Assumptions
Most researchers seem to be making individual assumptions on
what they consider “fame” or “success”. Simonton’s papers and
papers citing him often use his denitions of fame and melodic
originality, but many papers have dierent wording. They often
assume that the relationship between melodic originality and the-
matic fame is a curvilinear relationship or inverted-U as Simonton
arXiv:2210.12201v1 [cs.MM] 21 Oct 2022
Hudson Griith
showed in his papers. Despite Simonton’s results some papers such
as Hass’s 2015 paper show results that contrast this. With his analy-
sis of 20th-century popular music Hass stated that the “relationship
between fame and originality is not linear” [
3
] when referring to his
gure. This raises questions on whether or not the relationship be-
tween melodic originality and thematic fame Simonton discovered
only applies to the classical period he analyzed. With my research
topic, I plan to analyze a more specic period of classical music
which should answer this question.
2.3 Popular Methodologies
When it comes to the types of studies done in this eld, all pa-
pers utilize or focus on computational content-analyses. The main
methodology in this eld is running the transcribed notes through
algorithms to nd the coecients for variables such as melodic
originality, aesthetic success, and originality variation. More recent
papers such as the work done by Christof Weiß [
20
] and Lesley
Mearns [
6
] utilize MIDI les, which had yet to be invented dur-
ing Simonton’s research. While MIDI does have it downsides like
the ones mentioned in "The Dysfunctions of MIDI" [
8
], it is still
widely used in the eld of music and very useful for modern com-
putational analyses. However, there is an alternative to analyzing
notated music les, which is signal analysis. Signal analysis, or sig-
nal processing as it is also known, is the analysis of the actual audio
of a song recording. According to research by Meinard Müller, a
professor for Semantic Audio Processing at the International Audio
Laboratories Erlangen in Germany, music-based signal analysis is
a growing eld thanks to advances in speech signal processing;
however, it is often more complex and dicult than the analyzing
notated les [
9
]. Signal analysis has been used to attempt to predict
musical popularity in research by Junghyuk Lee [
5
] as well as the
emotion created by a musical piece in research done by Yazhong
Feng [
2
]; furthermore, signal analysis has yet to be used to detect
melodic originality. This is most likely because the popular method
used to calculate melodic originality requires the specic notes of a
piece which is often dicult for signal analysis to detect perfectly.
When it comes to programming languages, many papers do not
state the specic language used; however, one paper by Hass [
3
]
states that he used R and even provides information on where to
nd the actual code he used. His code provided is an important ref-
erence when conrming the validity and testing the code I created
for this analysis. Many sources do provide the specic formulas
used to calculate the dierent variables which again helps conrm
my methodology’s validity.
2.4 Common Findings and Conclusions
The most important nding in the eld of analyzing melodic origi-
nality and thematic fame is that they have a curvilinear relationship.
Figure 1: Relationship between melodic originality and ob-
jective repertoire popularity [1].
Figure 1 shows this relationship and describes "how a theme’s
melodic originality (as computed from two-note transition probabil-
ities) is associated with three musical criteria: objective repertoire
popularity (a composite measure of the performance and record-
ing frequency of the composition which contains the theme) and
subjective assessments of its aesthetic signicance and listener
accessibility" [1].
This was discovered by Simonton in 1980 and is the most com-
monly cited conclusion when it comes to the relationship between
melodic originality and popularity; however, there is research that
does not match this conclusion. An example of this would be the
previously stated research by Richard Hass and his nding that
the “relationship between fame and originality is not linear” [
3
].
When it comes to variables besides popularity, popular ndings are
that melodic originality increases with age and is greater for music
composed during periods of biographical stress [14].
2.5 Gap in Research
The gap in research that will be investigated in this paper is a new
approach to calculating melodic originality. Below is the formula
Simonton used to calculate melodic originality [15].
𝑂𝑟𝑖𝑔𝑖𝑛𝑎𝑙𝑖𝑡𝑦 =1− ( Í1<𝑖<5𝑃𝑖
5)
This can be simplied as the sum of all two-note transition probabil-
ities for the rst 5 bigrams, P(i), divided by 5. This gives the average
two-note transition probability over the rst 6 notes. That is then
subtracted from 1 to give the improbability over the rst 6 notes or
how unlikely the rst 6 notes are in a musical piece. There is a gap
in this because it only analyzes the rst 6 notes. The novel method
introduced in this paper analyzes all notes in a piece providing
摘要：
展开>>
收起<<
AcomputationalanalysisontherelationshipbetweenmelodicoriginalityandthematicfameinclassicalmusicfromtheRomanticperiod.HudsonGriffithgriffithh@ufl.edu1INTRODUCTIONThroughoutmodernsociety,manyindividualshaveattemptedtopredictpopularity.Whetheritbethepopularityofacompany,anewproduct,oratweet.Thisisoften...
声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！
分类：图书资源
价格：10玖币
属性：7 页
大小：1.01MB
格式：PDF
时间：2025-04-30
渝公网安备50010702506394
