PreprintPDF Available

Why Are Song Lyrics becoming Simpler: A Time Series Analysis of Lyrical Complexity in Six Decades of American Popular Music

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Song lyrics are rich in meaning. In recent years, the lyrical content of popular songs has been used as an index of culture’s shifting norms, affect, and values. One particular, newly uncovered, trend is that lyrics of popular songs have become increasingly simple over time. Why might this be? Here, we test the idea that increasing lyrical simplicity is accompanied by a widening array of novel song choices. We do so by using six decades (1958-2016) of popular music in the United States (N = 14,661 songs), controlling for multiple well-studied ecological and cultural factors plausibly linked to shifts in lyrical simplicity (e.g., resource availability, pathogen prevalence, rising individualism). In years when more novel song choices were produced, the average lyrical simplicity of the songs entering U.S. billboard charts was greater. This cross-temporal relationship was robust when controlling for a range of cultural and ecological factors and employing multiverse analyses to control for potentially confounding influence of temporal autocorrelation. Finally, simpler songs entering the charts were more successful, reaching higher chart positions, especially in years when more novel songs were produced. The present results suggest that cultural transmission depends on the amount of novel choices in the information landscape.
Content may be subject to copyright.
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Why are song lyrics becoming simpler? A time series analysis of lyrical complexity in six
decades of American popular music
Michael E. W. Varnum1, Jaimie Arona Krems2, Colin Morris3, Alexandra Wormley1, Igor
Grossmann4
1Arizona State University, 2Oklahoma State University, 3Toronto, Canada, 4University of
Waterloo
Correspondence to:
1Michael Varnum
mvarnum@asu.edu
2Jaimie Arona Krems
jaimie.krems@okstate.edu
3Colin Morris
colin.morris2@gmail.com
4Igor Grossmann
igrossma@uwaterloo.ca
Word Count: 4810 Words (excluding references and abstract)
Abstract: 187 Words
1
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Abstract
Song lyrics are rich in meaning. In recent years, the lyrical content of popular songs has been
used as an index of culture’s shifting norms, affect, and values. One particular, newly uncovered,
trend is that lyrics of popular songs have become increasingly simple over time. Why might this
be? Here, we test the idea that increasing lyrical simplicity is accompanied by a widening array
of novel song choices. We do so by using six decades (1958-2016) of popular music in the
United States (N = 14,661 songs), controlling for multiple well-studied ecological and cultural
factors plausibly linked to shifts in lyrical simplicity (e.g., resource availability, pathogen
prevalence, rising individualism). In years when more novel song choices were produced, the
average lyrical simplicity of the songs entering U.S. billboard charts was greater. This cross-
temporal relationship was robust when controlling for a range of cultural and ecological factors
and employing multiverse analyses to control for potentially confounding influence of temporal
autocorrelation. Finally, simpler songs entering the charts were more successful, reaching higher
chart positions, especially in years when more novel songs were produced. The present results
suggest that cultural transmission depends on the amount of novel choices in the information
landscape.
2
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Why are song lyrics becoming simpler? A time series analysis of lyrical complexity in six
decades of American popular music
Music is a human universal (Dissanayacke, 2000; Mehr et al., 2019), and it is known to
influence cognition, affect, and behavior (e.g., Anderson et al., 2003; Krumhansl, 2002; DeWall
et al., 2011). Because songs—and particularly popular song lyrics—can be so rich in meaning
(Cooper, 1985; Hayakawa, 1957), social scientists have long explored the ways that such lyrics
intersect with some fundamental social processes, including identity formation and person
perception (e.g., Hyden & McCandless, 1983; Marshall & Naumann, 2018; Reisman, 1957;
Rentfrow & Gosling, 2003, 2006, 2007).
More recently, social psychologists have begun to view music as a cultural product and to
examine the ways that popular music lyrics reflect important aspects of psychology at the
cultural level; the content in popular lyrics indexes changing norms, affect, and/or values (e.g.,
Brand et al., 2019; DeWall et al., 2011; Diamond et al., 2006; Eastman et al., 2015; Lambert et
al., 2020; Pettijohn & Sacco, 2009a, 2009b). For example, DeWall and colleagues (2011)
explored popular song lyrics as a “window into understanding U.S. cultural changes in
psychological states” (p. 200), finding that popular songs lyrics from 1980-2007 reflected an
increase in self-focus and a decrease in other-focus.
Here, we demonstrate that popular music lyrics have become increasingly simple over
time, and we test one possible explanation for this surprising trend, namely that the amount of
novel song choices has increased.
Novel Song Choices and Lyrical Simplicity
3
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Several lines of evidence suggest that people may have baseline preferences for songs
with simpler lyrics. One of the most widely known phenomena in psychology is the mere
exposure effect, a phenomenon where repeated exposure to a non-aversive stimulus increases
preference for it (Bornstein, 1989; Zajonc, 1968, 2001). One implication of this principle for the
present question is that simpler, more repetitive lyrics as these pieces essentially have this effect
baked into them and thus may tend to be preferred all other things being equal. Further, songs
with more repetitive lyrics may enjoy certain advantages in terms of information transmission as
they are easier to remember (Margulis, 2014) and likely easier to transmit with fidelity (Bartlett,
1932; Henrich, 2015; Rubin, 1997). Further, recent work has shown that naïve listeners find
simpler, more repetitive pieces of music to be more enjoyable, engaging, and memorable
(Margulis, 2013, 2014).
Why might pop songs become lyrically simpler in times when more new songs are
produced? Theory and research from diverse literatures suggest that songs with simpler lyrics
might be especially successful when there are more new songs to choose from. First, humans are
cognitive misers. People have limited information-processing capacities (e.g., Cowan, 2001), and
are known to conserve mental resources (Fiske & Taylor, 2013). Consequently, humans often use
shortcuts in decision-making (Bargh, 1989; Tversky & Kahneman, 1974). For example, when
confronted with the task of evaluating persuasive messages and/or complex decision
environments, people are more likely to use heuristics, peripheral cues, and other automatic
cognitive processes to evaluate these messages if cognitive resources are limited in some fashion
(Eagly & Chaiken, 1993; Petty & Cacioppo, 1986). Thus, when there are more products to be
evaluated, people may increasingly prefer simpler products as they may require less mental effort
to engage with. The mere exposure effect might also have a greater influence on decision making
4
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
in such contexts as well, given that it too can be thought of as a heuristic or even instinctive
evaluation. Further, across real-world studies and in-laboratory experiments, when people are
confronted with a greater number of options to choose from, they are more likely to choose
simpler, less cognitively demanding products (Iyengar & Kamenica, 2010). Taken together, this
work suggests that pop songs on average might become lyrical simpler in times when people are
exposed to greater amounts of new songs and that success of such songs might be more strongly
linked to lyrical simplicity in such times.
Here, we test the hypothesis that the trend toward increasingly simple popular music
lyrics might be accompanied by the increasing number of songs released each year, using six
decades’ worth of song data. We also do so while including a number cultural and ecological
control variables, as prior work demonstrates that well-studied ecological features, such as
resource levels, pathogen threat, and sources of external threat (e.g., climatic stress, armed
conflict) can impact markers of cognition and behavior at the cultural-level (Jackson et al., 2019;
Santos et al., 2017; Sng et al., 2018; Varnum & Grossmann, 2017), and might plausibly affect
preferences for simplicity in aesthetic products. For example, both resource scarcity and
pathogen prevalence have been associated with conformity, innovation, and creativity in prior
work (e.g., Jackson et al., 2019; Stephens et al., 2007; Varnum & Grossmann, 2019).
Methods
We gathered cross-temporal data covering a period of six decades (1958-2016) on lyrical
compressibility (as an index of simplicity/complexity of song lyrics), amount of novel songs
produced (as an index of available novel song choices), and ecological, socioecological, and
cultural variables linked to patterns of cultural change in previous research or plausibly related to
trends in aesthetic content.
5
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Lyrical compressibility of successful music. We gathered data from 14,661 songs that
entered the Billboard Hot 100 charts spanning the period from 1958 (the charts inception) to
2016. The Billboard Hot 100 tracks the 100 most popular songs each week based on music sales,
radio airplay, and internet streaming. To operationalize lyrical complexity (vs. simplicity), we
estimated text compressibility.1
Compressibility indexes the degree to which song’s lyrics have more repetitive and less
information dense, and thus simpler, content. We used a variant of the established LZ77
compression algorithm. In brief, the LZ77 algorithm works by finding repeated substrings and
replacing them with 'match' objects pointing back to the string's previous occurrence. A match is
encoded as a tuple (D, L), with D being the distance to the substring's previous occurrence, and L
being its length. We treated these matches as costing 3 bytes. This way, a repeated string only
leads to space savings if it is of at least length 4, and longer repetitions lead to greater relative
savings. Given a song S, and the set of matches M produced by the LZ77 algorithm when applied
to that song, its compressed size is therefore:
1 By operationalizing complexity via a compressibility index, we avoided some of the
conceptual ambiguity associated with operationalization of complexity in prior research (e.g.,
Jordan et al., 2018; Kempe et al., 2004; Rycroft, 2006): Whereas multi-purpose use of a single
product may reflect product’s complexity from the operational standpoint, it may also represent
greater simplicity from the standpoint of consumer psychology. Further, song lyrics are tractable
to work with when using an automated compression algorithm.
6
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Where
|
S
|
is the original size of the song's lyrics, measured in characters/bytes. The
compression ratios of songs in our dataset (i.e., |S|/compsize(S)) followed an approximately log-
normal distribution, so we operationalized compressibility as the logarithm of this ratio:
We used the LZ77 compression algorithm because of its intimate connection to textual
repetition. Most of the byte savings when compressing song lyrics arise from large, multi-line
sections (most importantly the chorus, and chorus-like hooks). Another significant contributor
are multi-word phrases, which may be repeated in variations across different lines for poetic
effect (e.g. the anaphoric verses in Lady Gaga's Bad Romance: "I want your ugly / I want your
disease / I want your everything..."). The compression may make use of repeated individual
words, or even sub-word units that repeat (perhaps incidentally), but their contribution to the
overall compressibility is low.
Higher compression scores signify more repetition and therefore higher simplicity. A
score of 0 means no compression was possible (e.g. if the input were random noise), a score of 1
means a 50% reduction in size, a score of 2 means a 75% reduction, and so on. For example,
Daft Punk’s 1997 song “Around the World” repeats the title 144 times and has a compressibility
score of 5.42 (the maximum in this sample). Nat King Cole’s “The Christmas Song” (1961) has a
low compression score of 0.11.
We computed mean compressibility for each year based on all songs that entered the Hot
100 charts in a given year for which we were able to scrape lyrics (1958-2016). Because we used
an automated procedure for song scraping, which depends on the readability of the song lyrics,
the percentage of songs scraped varied between 27% of top 100 songs in 1958 and 91% of songs
in 2015 (M = 57%, Md = 57%, SD = 19%). Because percentage of scraped songs has been
7
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
increasing over time, and correlated with the compressibility index, τ = .73, p < .001, in
additional analyses we controlled for this trend.
Song success. Some of the theoretical positions we draw on to evaluate possible reasons
for changes in lyrical complexity suggest that more compressible songs may be more likely to be
successful. To evaluate this proposition, we additionally gathered data on the highest position of
each song in the sample achieved on the Billboard charts.
Novel music production. In the spirit of the multiverse analyses (Steegen et al., 2016),
we used three separate indicators to assess the amount of new music to which people are likely
exposed in a given year. For each year (1958-2016) we computed the total number of songs
which made the Hot100 chart, the number of musical releases per year according to Discogs
(Discogs.com), and the number of Wikipedia entries about songs first published or performed
each year (Wikipedia.org).
Possible ecological drivers of cultural change in aesthetic preferences and music
production. We assessed a range of well-studied socioecological factors (e.g., resource levels,
pathogen threat, sources of external threat), which could plausibly bear on aesthetic preferences
or might affect lyrical simplicity (and whether the predicted association between novel music
production and simplicity holds even controlling for these or other ecological and cultural
variables discussed below). Resource scarcity has been linked to greater conformity (Stephens et
al., 2007) and cross-temporal work has found that greater resource levels are linked to more
innovation and creative output (Varnum & Grossmann, 2019) and less conformity (Bianchi,
2016; Grossmann & Varnum, 2015). Higher levels of infectious disease have also been linked to
more conformity (Murray, Trudeau, & Schaller, 2011; Horita & Takezawa, 2017), traditionalism
(Tybur et al., 2016), and tight social norms (Gelfand et al., 2011; Jackson et al., 2019). External
8
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
threats, due to climate or war, have also been linked to more traditional outlooks and tight social
norms (Gelfand et al., 2011), which might similarly bear on trends in lyrical simplicity. We thus
included publically accessible data indexing these factors GDP per capita, GDP growth,
unemployment, pathogen prevalence, climatic stress, and participation of the US in major armed
conflicts. The data used in our analyses covered the years 1958-2016. Data on GDP per capita
and GDP growth were gathered from macrotrends.net, and data on the other markers came from
Varnum & Grossmann (2016) and updates from the original data sources used in that publication.
We also explore the possible impact of other socioecological factors that might plausibly
affect lyrical simplicity. One might speculate that immigration could drive increases in lyrical
simplicity. For example, simpler lyrics in American pop songs might be linked to shifts in the
amount of people for whom English may not be a first language. In a similar way, it might be
that ethnic fractionalization, so far linked to changes in individualism and uniqueness over time
(Huynh & Grossmann, in press)¸may also increase preferences for, memory of, and/or dispersal
of simpler, more repetitive lyrics, as such content would be easier to convey and understand to a
wide range of audiences. To assess the possibility that a rise in simpler English lyrics might be
linked to shifts in the amount of people for whom English may not be a first language, we used
data on the number of green cards issued from the Department of Homeland Security as a marker
of immigration. To assess possibilities linked to ethnic fractionalization, we used data on ethnic
fractionalization from the US Census Bureau.
Research on the consequences of residential mobility also suggests that perhaps this
variable might also affect lyrical trends. Previous studies have linked residential mobility to
greater susceptibility to the mere exposure effect and greater preference for familiar cultural
products (Oishi et al., 2012); thus, it may be that mobility is also linked to temporal variations in
9
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
lyrical complexity of pop songs. To assess residential mobility, we gathered data on percentage
of the US population that changed residence within the US from the US Census Bureau.
At the same time, a simpler variable might also be driving this effect. Perhaps products
that succeed with a larger audience are merely simpler, akin to a lowest common denominator
effect. Because the U.S. population grew substantially in recent decades, we also test whether
population trends might be associated with lyrical simplicity. Thus, we also gathered data on the
total size of the US population from macrotrends.net to explore population size.
Cultural factors. Prior work has found conservatives show a preference for simple and
unambiguous art, speech patterns, and literature (Wilson et al., 1973; Schoonvelde et al., 2019;
Suedfeld, 2010; McAllister & Anderson, 1991; Jost et al., 2003 though see also Conway et al.,
2016). Thus, one might suspect that possible changes in conservatism could be driving lyrical
simplicity. Somewhat similarly, other evidence suggests that cross-cultural differences in
aesthetic preferences and expression are linked to orientations toward collectivism (Morling &
Lamareux, 2008; Nand et al., 2014). Thus, we also gathered data on indicators of conservative
ideology, operationalized conservatism as the average percent of annual survey respondents in
Gallup polls identifying as conservative, and we included as an index of cultural level
collectivism based on frequency of collectivism related words in the Google Ngrams American
English corpus (Grossmann & Varnum, 2015).
Analytic procedure. Where possible, we use non-parametric ordinal-level measures of
correlation or partial correlation (Kendall’s rank correlation coefficient τ), which provides
estimate of similarity of the orderings of the data when ranked by each of the quantities. Since
Fechner’s initial work on time series analyses, Kendall’s τ has been a preferred metric for
examining cross-temporal relationships (Kruskal, 1958). It provides a conservative estimate,
10
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
which is preferred because time series data is rarely normally distributed. Results were
comparable when we used Pearson’s r or partial Pearson correlations. In the initial step, we
examined zero-order relationships between each of the three indices of available novel song
choices and average lyrical compressibility of popular songs. Next, we created a composite index
of novel song choices and assessed the robustness of the hypothesized link between amount of
novel song choices and average lyrical compressibility of popular songs by controlling for a host
of ecological, socioecological, and cultural factors that might plausibly influence cultural level
success for simplicity vs. complexity. Our chief analyses focused on a set of corrective analyses,
in which we controlled for the possibly spurious nature of the relationship between our key time
series due to temporal autocorrelation.
Given the range of possibilities of correcting for temporal autocorrelation, we opted to
perform three different types of analyses that correct or account for the possibility that observed
relationships might be spurious as a function of autocorrelation in the time series. First, we
computed adjusted significance thresholds based on the Tiokhin-Hruschka procedure (Tiokhin &
Hruschka, 2017). Second, we detrended our novel song production and lyrical compressibility
time series by residualizing for year and assessed the correlation between our detrended
variables. Finally, for central univariate and multivariate analyses, we used an automated auto-
regressive integrated moving average forecasting model (auto.ARIMA) to assess the relationship
between novel song choices and lyrical compressibility (Hyndman & Khandakar, 2008). This
technique involves a machine learning algorithm that tests a number of different possible models
which vary in autoregressive components, differencing, and moving average components, as well
as whether they include an exogenous predictor. Additionally, we used auto.ARIMA to generate
a forecast for future patterns of lyrical compressibility (2017-2046).
11
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
For multivariate analyses we entered multiple predictors of lyrical compressibility over
time. To avoid multicollinearity and overfitting (and due to limited number of units at the yearly
level of analysis), we first aggregated covariance scores attributed to additional socioecological
and cultural factors (see Table 1) by performing a principal component analysis on these
covariates and saving component scores for further multivariate time series analyses. The first
principal component explained 50% of the variance in the covariates, with strong loadings
(absolute value >.85) for Population Size, GDP/capita, Residential Mobility, Pathogen
Prevalence, Ethnic Heterogeneity and Immigration, moderate loadings for Armed Conflicts (.49)
and weak loading of GDP growth (.44). Other covariates (Climatic Stress, Unemployment,
Conservatism, Collectivism) showed very weak loadings (.21 < absolute value .27). Next, we
entered both yearly music production scores and covariate-PCA scores as independent predictors
of lyrical compressibility, simultaneously accounting for the time series structure in the data.
Data availability. All data and reproducible code for analyses reported in the manuscript
are available on the Open Science Framework (https://osf.io/qnsmj/).
Results
Indicators of novel song choices and average lyrical compressibility. As Figure 1
indicates, mean lyrical compressibility (i.e., simplicity) of songs increased over time, Kendall’s τ
= .726, p < .001, as did number of songs making the Hot 100 charts per year, Kendall’s τ = .425,
p < .001, number of music releases according to Discogs per year, Kendall’s τ = .973, p < .001,
and number of Wikipedia entries for songs by year of publication, Kendall’s τ = .871, p < .001.
Figure 1. Change in lyrical compressibility, along with a music production-based forecast for
future lyrical compressibility from regression with ARMIA (1,0,0) and index of novel song
12
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
choices as an exogenous predictor. Light purple indicates 95% confidence bands, dark purple
indicates 80% confidence bands.
Analyses of the composite index of novel song choices. Hot100 songs, Discog music
releases, and Wikipedia song entries were highly correlated, .41 < Kendall’s τ’s ≤ .87, and
formed a single principle component with highest loadings by the Wikipedia song entries (.98),
and weakest loading by the Hot 100 songs (.88). To avoid multicollinearity, we used component
scores for further analyses. Overall, this index of novel music production was strongly positively
related to compressibility, Kendalls τ = .714¸ p < .001.2
2 Consistent with our predictions, mean lyrical compressibility per year was positively correlated with amount of
novel music produced per year as operationalized by three distinct indicators, Kendall’s τ (n songs in Hot 100
charts/year) = .429, p < .001, Kendall’s τ (n Discogs music releases / year) = .721, p < .001, Kendall’s τ (n Wikipedia
entries about songs/year) = .680, p < .001.
13
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Relationships between socioecological factors and compressibility. Although several
ecological dimensions were associated with changes in average lyrical compressibility over time
(see Table 1), these relationships were often in the opposite direction that prior research or
theorizing would suggest. For example, there were significant negative correlations between
GDP per capita and pathogen prevalence and average lyrical compressibility. Further, our two
cultural variables were either unrelated to lyrical compressibility (conservatism) or correlated in
the opposite of the predicted direction (collectivism). We did observe theoretically sensible
relationships between compressibility and residential mobility, immigration, ethnic
fractionalization, and population size. However, when controlling for the potentially confounding
effect of temporal auto-correlation by residualizing out the effect of year, only three of these
relationships are statistically significant, and only the relationship between pathogen prevalence
and average lyrical complexity remains in a theoretically sensible direction (see Table 1).
Robustness analyses: Control variables. This PCA-based composite index of music
production remained significantly related to lyrical compressibility when including percentage of
scraped songs/year as a covariate, Kendall’s τp = .261¸ p = .003. Further, it remained significant
when controlling separately for each of the 12 specified control variables, .220 < partial
Kendall’s τ’s < .770, p’s < .02 (see Table 2 for details). Full correlations between these variables
are presented in Figure S1.
Table 1. Correlations with average lyrical compressibility.
Variable Kendall’s
τ
Kendall’s τ
(Detrended)
14
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
INFORMATION
LANDSCAPE
Music Production
.714***
.222**
ECOLOGICAL
GDP per capita .733*** .044
GDP growth -.260** -.073
Unemployment .051 .135
Pathogen Prevalence -.490*** .324**
Climatic Stress -.118 .050
Armed Conflict .229* .063
SOCIO-ECOLOGICAL
Immigration .563*** -.155
Ethnic Heterogeneity .737*** -.066
Residential Mobility -.692*** -.230*
Population Size .726*** .135
CULTURAL
Conservatism -.019 -.287*
Collectivism -.225* -.124
*p < .05, ** p ≤ .01, *** p ≤ .001
Table 2. Partial correlations between novel music production index and average lyrical
compressibility.
15
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Control Variable Partial Kendall’s τ
Novel Music
Production & Lyrical
Compressibility
ECOLOGICAL
GDP per capita .248**
GDP growth .695***
Unemployment .753***
Pathogen Prevalence .596***
Climatic Stress .710***
Armed Conflict .696***
SOCIO-ECOLOGICAL
Immigration .539***
Ethnic Heterogeneity .267**
Residential Mobility .436***
Population Size .231**
CULTURAL
Conservatism .670***
Collectivism .610***
*p < .05, ** p ≤ .01, *** p ≤ .001
Robustness analyses: Auto-correlation. Importantly, the correlation between this
composite index of novel song choices and average lyrical compressibility remained significant
when adjusting significance thresholds using the Tiokhin-Hruschka method to account for
observed auto-correlation in the two time series, r = .877¸ correctedp < .001. As an alternative
method for dealing with autocorrelation, we also detrended the time series by residualizing out
the linear impact of year. The correlation for our detrended variables remained significant,
Kendall’s τ = .222, p = .010.
Given the time series nature of our data, another way to test the hypothesized link
between amount of new songs available and average compressibility of these songs while also
16
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
addressing the issue of autocorrelation can involve an automated ARIMA algorithm
(auto.ARIMA) within the forecast package (Hyndman, et al., 2020) in R 4.0.0 (R Core Team,
2020). This machine-learning algorithm inspects the time-series data to fit the optimal
forecasting function. The auto-regressive (AR(p)) component refers to the use of past values in
the regression equation for the series Y. The auto-regressive parameter p specifies the number of
lags used in the model. A moving average (MA(q)) component represents the error of the model
as a combination of previous error terms et. The order q determines the number of terms to
include in the model. ARIMA models are well-suited for long-term time series, such as the
historic patterns in the present data. The automated algorithm within the forecast package
searches through combinations of order parameters and picks the set that optimizes model fit
criteria, comparing Akaike information criteria (AIC) or Bayesian information criteria (BIC) of
respective models. Notably, the automated forecasting approach allows us to specify an
exogenous predictor such as novel song choices, such that the automated function can evaluate
the extent to which this exogenous predictor improves the fit above and beyond the
decomposition of the time-series of the dependent variable. In other words, the automated
function provides a conservative way to see whether an exogenous predictor such as the novel
song choices index improves accuracy in forecasts of the lyrical compressibility. If the final
model selected by auto.ARIMA includes our putative exogenous variable (in this case amount of
novel song choices), then this suggests that this variable helps the model to achieve optimal fit to
the data.
The results of this automated forecasting procedure indicated that a model with a
positive autoregressive component, B = .527, SE = .124, and a positive contribution of the novel
music production index, B = .059, SE = .008, provides the best fit to the data:
17
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
yt (lyrical compressibility function) = .983 + .527yt-1 + .059x + et
This model estimation suggests that the index of novel song choices contributes to average
lyrical compressibility above and beyond the temporal autocorrelation observed for average
lyrical compressibility. Further, the coefficient for the index of novel song choices was
statistically significant, z = 6.95, p < .001.
We also ran an alternative set of auto.ARIMA analyses where we set novel song choices
as the dependent variable and average lyrical compressibility as an exogenous predictor. The
results of this automated forecasting procedure indicated that a model with two positive moving
average components, B = 1.176, SE = .242, and B = .487, SE = .164, and a positive contribution
of average lyrical compressibility, B = 5.067, SE = 2.207, provides the best fit to the data:
yt (novel music production function) = - 4.991 + 1.176εt-1 + 0.487εt-2+5.067x + et
The coefficient for lyrical compressibility was statistically significant, z = 2.30, p = .02.
Comparison of the Akaike Information Criterion (AIC) and Bayesian Information Criterion
(BIC) values for our primary and alternative models suggest that our primary model with novel
song choices as an exogenous predictor and lyrical compressibility as the dependent variable,
AIC = -235.84, BIC = -227.53, is superior to the alternate model with lyrical compressibility as
an exogenous predictor and novel song choices as the dependent variable, AIC = 58.36, BIC =
68.75.
Robustness Analyses: Controlling for percentage of scraped songs. Because of a
positive association between lyrical compressibility and percentage of scraped songs per year, we
performed a separate set of analyses in which we first regressed out the effect of sampling (% of
scraped songs/year) on lyrical compressibility and performed an auto.ARIMA analysis on the
residuals. Results of a model on the residuals with music production as a predictor indicated a
18
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
significant effect of music production, B = .799, SE = 0.046, z = 17.32, p < .001, suggesting that
the effect songs even when accounting for the possible change in sampling.
Multivariate analyses. In another set of control analyses, we performed an auto.ARIMA
analysis, in which we included the PCA factor formed by all socio-ecological covariates as a
second covariates. By comparing the magnitude of the effect from this first principal component
(which was chiefly driven by ecological variables) and music production index, we can assess
the relative contribution of the music production index via-a-vis other socio-ecological
covariates. The results of this automated forecasting procedure indicated that a model with a
positive autoregressive component, B = .513, SE = .118, a significant positive contribution of the
novel music production index, B = .038, SE = .016, z = 2.37, p = .018, and a non-significant
positive trend formed by ecological covariates (and chiefly reflecting economic and population
growth), B = .026, SE = .016, z = 1.61, p = .108, provides the best fit to the data:
yt = .981 + .513yt-1 + .038(music production) + .026 (ecological covariates)+ et
This model estimation suggests that the index of novel song choices contributes to
average lyrical compressibility above and beyond the temporal autocorrelation as well as other
ecological covariates observed for average lyrical compressibility. Moreover, the effect of music
production on lyrical compressibility was stronger than other feasible covariates explored in the
present dataset.
Exploratory song-level analyses. In exploratory analyses we evaluated how lyrical
compressibility is associated with song success, and whether this relationship was stronger in
time periods when more novel music was produced. Given that we shifted focus to song-specific
data, we utilized a multi-level framework via lme4 package in R, with songs’ chart position and
lyrical compressibility scores nested within years. Preliminary auto.ARIMA analyses on the
19
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
yearly aggregate data indicated that a model with no auto-regressive components but a linear
trend would show the best model fit. Therefore, in the first multi-level model we included year as
a proxy for a linear trend as well as compressibility X year interaction as predictors of song
success. Both year and lyrical compressibility were mean-centered prior to analyses. This multi-
level model showed a good overall model fit, R2 = .05, with 3.9% of the variance explained by
fixed effects. Results indicated a significant effect of year, B = 0.318, SE = 0.031, t(df = 57.29) =
10.23, p < 001, suggesting that over time songs included in the sample on average had a lower
chart rank—a typical regression to the mean effect. Importantly, more compressible songs
showed significantly higher rank in the charts, B = - 9.321, SE = 0.661, t(df = 14640.88) = 14.10,
p < .001, and this effect was particularly pronounced for more recent years, compressibility X
Year interaction, B = - 0.105, SE = 0.039, t(df = 14581.41) = 2.71, p = .007.
In the second step, we added mean-centered yearly music production index as a second
covariate, along with a music production X compressibility interaction. Based on prior
auto.ARIMA results, we also included linear effect of year to account for the trend in the chart
position. This multi-level model also showed a good overall model fit, R2 = .06, with 4.7% of the
variance explained by fixed effects. More compressible songs showed significantly higher rank
in the charts, B = - 9.353, SE = 0.657, t(df = 14819.95) = 14.23, p < .001. Also, average chart
position of songs was higher in years with a greater volume of songs produced, B = 6.141, SE =
1.280, t(df = 53.76) = 4.80, p < .001. Moreover, as Figure 2 indicates, lyrical compressibility
was more strongly associated with song success in years with greater volume of produced songs,
compressibility X music production interaction, B = - 2.170, SE = 0.648, t(df = 14781.15) = 3.35,
p = .001. These analyses yield results consistent with the proposition that lyrically simpler songs
enjoy greater success in time periods in which more novel song choices are available.
20
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Figure 2. Relationship between lyrical compressibility and chart position for years differing in
music production volume. Confidence bands indicate 95% around the estimate.
Forecasting. As a final step, we generated a forecast for average lyrical compressibility
for four decades after the last data point in our time series. This is in keeping with
recommendations by Varnum & Grossmann (2017) that papers analyzing past patterns of cultural
change provide forecasts for the future. These forecasts enable a test of this theoretical model
against concrete future cultural trends. Using the automated ARIMA algorithm, we also
identified the best function for the novel song choices data, which we used to estimate the
subsequent 40 data points. In turn, we used this estimated data in conjunction with the
21
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
compressibility function to forecast the further development of lyrical compressibility. Results of
this model suggest that lyrical compressibility will continue to increase over the next several
decades (see Figure 1).
Discussion
Popular music lyrics have recently been used to inform work on the cultural transmission
of emotional expression (Brand et al., 2019; Schellenberg & von Scheve, 2012), as an index of
culture-level changes in self- versus other-focus (DeWall et al., 2011), and as a reflection of
cultural mood in respond to economic and social threats (e.g., Pettijohn & Sacco, 2009a, 2009b).
But one major trend in popular music lyrics remained underexplored and unexplained—popular
music lyrics are coming increasingly simple over time. We reasoned and found support for the
hypothesis that increasing lyrical simplicity is associated with increasing amounts of novel music
production. That is, in times when more novel music is produced, popular songs become
increasingly lyrically simple.
The relationship between mean lyrical compressibility and the amount of novel music
produced each year was robust. We observed significant positive associations across three
operationalizations of the amount of novel song choices and the average lyrical compressibility
of popular songs. Further, the relationship between amount of novel song choices and average
compressibility of popular songs remained significant when including a host of ecological,
socioecological, and cultural factors linked to other types of cultural change both in univariate
and multivariate analyses. By and large these other variables were not significantly associated
with changes in lyrical simplicity after controlling for the potentially confounding influence of
temporal autocorrelation. Of note, we also observed a significant negative association between
changes in pathogen prevalence and lyrical simplicity. This observation suggests a potentially
22
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
new consequence of infectious disease threat, one that should be explored in more detail in future
work.
Importantly, the linkage between amount of new music produced and average
compressibility of popular songs also held when accounting for temporal autocorrelation using
three distinct methods. Thus, results suggest that the amount of novel music produced contributes
to changes in average lyrical compressibility above and beyond other plausible causes and
autoregressive trends in the data.
In exploratory analyses, we also found evidence suggesting that success, as indexed by
position in the billboard charts, among popular songs was associated with greater lyrical
compressibility. This is broadly consistent with the notion that simpler content enjoys an
advantage in memorability and/or transmission. Importantly, this effect appeared to be stronger
in years when the amount of novel songs produced was higher, providing conceptual
confirmation of our key finding. More novel song choices appear linked to both greater average
lyrical compressibility of the body of songs that succeeds (i.e., those entering the billboard chart
in a given year), and, among songs entering the charts in a given year, compressibility was more
strongly associated with better performance on the chart in years when more novel songs were
produced.
This finding might parallel ongoing research taking information-theoretic approaches in
exploring communicative efficiency in human language (Gibson et al., 2019; Tooby &
Cosmides, 2020). For example, in both language and music, something akin to Zipf’s law seems
to be at play (e.g., Mehr et al., 2019)—i.e., the frequency rank of a phenomenon is inversely
proportional to its probability, such that, in the case of language, many words are quite rare, but a
few words (e.g., pronouns) appear with great frequency. Moreover, these more successful (i.e.,
23
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
frequently-used) words tend be shorter in length (but see also Piantadosi et al., 2011). This
observation dovetails with our finding regarding the success of simpler lyrics. Indeed, the
increasingly success of simple lyrics may reflect increasing communicative efficacy.
A preference for simpler information in increasingly information-saturated environments
might also be consistent with some propositions from cultural evolutionary theory. One tenet of
cumulative cultural evolutionary theory is that human innovation, transmission, and learning
increase the amount and quality of cultural information, while also increasing the learnability of
this information (e.g., Boyd & Richerson, 1988; Henrich, 2015). One way to increase
information learnability is via simplicity (e.g., Kirby, Cornish, & Smith, 2008; Kirby et al.,
2015), thereby yielding increasingly efficient communication.
The present report adds to two growing bodies of empirical research—work emphasizing
the examination of cultural products as a window into cultural-level psychological processes
(e.g., Brand et al., 2019; DeWall et al., 2011) and work using time-series methods to test
hypotheses regarding the causes of particular patterns of cultural change (for a review see
Varnum & Grossmann, 2017). Here, we use big data and time series methods to show that
increases in the amount of novel songs over time appear to be linked to the increasing simplicity
of popular songs’ lyrics, as well as greater success of songs with simpler lyrics. What does this
tell us more broadly about how American culture has changed? It suggests potentially that
success of aesthetic complexity at the cultural level may be something that shifts over time.
Although this is not the first such demonstration of this phenomenon, to our knowledge this is
the first attempt to formally evaluate why such cultural-level preferences may change.
Alternative and Complementary Explanations
24
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Although we found that our key effect was highly robust, alternative or complementary
explanations for the growing success of lyrically simpler songs are still possible. For example,
changes in the ways that people consume popular music could perhaps affect lyrical simplicity.
Technological innovation (e.g., various portable music devices) could play a role, as could other
variation in the ways that people interact with music. Relatedly, one might speculate that the
success of increasingly simple lyrics might owe to technologically mediated increases in
listening to music primarily in the background (e.g., on commutes, in gyms). However, one
might easily argue that for generations music has been consumed in this fashion albeit with
slightly different technologies—portable radios, car stereos, and portable music players have
existed and been widely used for decades. It would be interesting to attempt to assess this
question empirically, although we are not currently aware of high-quality time series data
relating to how and why people listen to popular music. Moreover, operationalization of these
indicators of technological innovations over time would be a potentially thorny problem. For
instance, what does it mean to own a Walkman in 1982 as compared to a similar device in 2002?
Nonetheless, it would be intriguing to assess these questions in future work.
Another possibility is that the length of songs may have changed over time affecting
average lyrical complexity. Thus, perhaps song lyrics are more compressible by virtue of songs
becoming shorter. However, a recent analysis of songs entering the Billboard charts over the
course of its history suggests, in fact, that the average song on the charts in the late 2010’s was
somewhat longer than those in the 1950’s and 1960’s, and similar in recent years to levels
observed in the 1970’s (Bannister, 2017). Thus, this alternative explanation cannot account for
the trends observed in the present analyses.
25
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
One might alternatively speculate that the rise in lyrical simplicity observed in the present
data might be related to trends in the popularity of different musical genres. Indeed, although this
is beyond the scope of the present work, it would be interesting to empirically assess how lyrical
complexity varies across popular music genres and whether trends within these genres over time
have been similar. Further, future work might assess whether the linkage between lyrical
simplicity and song success observed in our exploratory analyses varies within genres of popular
music or if genres that are on average simpler enjoy greater success in times of more music
production.
Limitations
It is worth noting that our analysis was restricted to a single type of cultural product. It
might be the case that empirical analysis of other domains might show similar trends and a
similar relationship between amount of novel content and success of simpler content, or it may
be that different dynamics are observed when considering television shows, videogames, or other
types of cultural products. For example, many have argued that television shows have become
more complex and intellectually stimulating in the past few decades, entering the so-called
“Golden Age of Television.” However, empirical work examining complexity over time in other
types of cultural products, including movies, news broadcasts, print newspapers, novels, and
political speech suggests that there is in fact a broad trend toward simpler content being
increasingly preferred, at least when it comes to the language used in these products (Jordan et
al., 2019). It is noteworthy that Jordan and colleagues (2019) used a different measure of
complexity, in this case use of a specific set of words indicate cognitive complexity, and that they
find that the strength of the decline in complexity varies across different types of cultural
products. Hence, future research may attempt to conceptually replicate our work by assessing
26
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
compressibility of other types of cultural products over time and whether the success of such
products is linked to the number of options or alternatives within that domain.
It is also worth noting that, in the present work, we assessed the simplicity of lyrics.
Songs might be complex or simple in other ways as well, in terms of rhythm, melody, number of
instruments played, and so on. Analyses of these features is beyond the scope of the present
work, but it would be interesting to see the extent to which similar or divergent patterns are
observed in these facets of successful popular music over time.
Our analysis was also limited to songs that were relatively successful over time—i.e.,
those that made the Billboard Hot 100 chart. This sample is quite large (N > 14,000), but it may
not be representative of all songs produced during this period. Further, we were able to
successfully scrape a greater proportion of more recent rather than older songs, which we
included in control analyses. Our sample captures a large chunk of popular music produced
during more than half a century and enables tests regarding linkages between novel music
choices, lyrical simplicity, and song success. A slightly different conceptual question may be
worthwhile addressing in future work: Does average complexity of all music produced change
along with shifts in the amount of music produced?
Our work is also limited by the fact that song success was operationalized by commercial
success in the US market. Although some cultural shifts in the past several decades appear to be
global in nature, such as rising individualism (Santos et al., 2017), this need not be the case for
all dimensions of culture. Different dynamics may potentially be observed in terms of song
success in parts of the world with different values, practices, and ecological conditions. Although
such an endeavor is beyond the scope of the present manuscript largely due to the lack of equally
27
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
rich time series data from other countries, it would be worthwhile to try to address this question
in the future.
Finally, the present work is limited by its correlational nature. Although our findings
appeared quite robust across different operationalizations of the independent variable—when
accounting for autocorrelation in various ways, and when controlling for a host of plausible
ecological, socioecological factors, and cultural values which have shifted over time—we cannot
completely rule out all alternative explanations for increasing success of songs with simpler
lyrics. Future work might attempt to quantify society level time series trends in conformity or
other biases linked to lyrical affect and music sampling (e.g., Brand et al., 2019; Youngblood,
2019), and assess whether the present findings hold when controlling for these variables as well.
Future work may also use in-lab methods to explore and disentangle the possible causal
mechanisms underlying the link between amount of novel song choices and success of songs
with simpler lyrics. For example, transmission chain methods (Mesoudi & Whiten, 2008) could
be employed to explore whether participants might find simpler lyrics more pleasing and
memorable when there is a greater number of other song-snippets competing for attention versus
when there is not.
Conclusion
Why have the lyrics of pop songs become simpler over time? Our findings suggest that
the answer may have to do with the proliferation of new songs available to consumers. The
present work represents one of the first attempts to use big data and time series methods to
quantify temporal shifts in information transmission dynamics at the societal level. Future work
may attempt to replicate and extend these findings into other types of complexity and other types
of cultural products.
28
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
References
29
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Anderson, C. A., Carnagey, N. L., & Eubanks, J. (2003). Exposure to violent media: The effects
of songs with violent lyrics on aggressive thoughts and feelings. Journal of Personality
and Social Psychology, 84(5), 960–971.
Bannister, M. (2020). The Billboard Hot 100: Exploring Six Decades of Number One Singles,
<URL: https://github.com/mspbannister/dand-p4-
billboard/blob/master/Billboard_analysis__100417_.md>.
Bargh, J. A. (1989). Conditional automaticity: Varieties of automatic influence in social
perception and cognition. Unintended Thought, 3, 51-69.
Bartlett, F. C. (1932). Remembering: An experimental and social study. Cambridge: Cambridge
University Press.
Bianchi, E. C. (2016). American individualism rises and falls with the economy: Cross-temporal
evidence that individualism declines when the economy falters. Journal of Personality
and Social Psychology, 111(4), 567.
Bornstein, R. F. (1989). Exposure and affect: overview and meta-analysis of research, 1968–
1987. Psychological Bulletin, 106(2), 265.
Boyd, R., & Richerson, P. J. (1988). Culture and the evolutionary process. University of Chicago
Press.
Brand, C. O., Acerbi, A., & Mesoudi, A. (2019). Cultural evolution of emotional expression in 50
years of song lyrics. Evolutionary Human Sciences, 1, E11.
Conway, L. G., Gornick, L. J., Houck, S. C., Anderson, C., Stockert, J., Sessoms, D., & McCue,
K. (2016). Are Conservatives Really More Simple-Minded than Liberals? The Domain
Specificity of Complex Thinking: Ideology and Complexity. Political Psychology, 37(6),
777–798.
Cooper, V. W. (1985). Women in popular music: A quantitative analysis of feminine images over
time. Sex roles, 13(9-10), 499-506.
Cowan, N. (2001). Metatheory of storage capacity limits. Behavioral and Brain Sciences, 24(1),
154-176.
DeWall, C. N., Pond Jr, R. S., Campbell, W. K., & Twenge, J. M. (2011). Tuning in to
psychological change: Linguistic markers of psychological traits and emotions over time
in popular US song lyrics. Psychology of Aesthetics, Creativity, and the Arts, 5(3), 200.
Diamond, S., Bermudez, R., & Schensul, J. (2006). What’s the rap about ecstasy? Popular music
lyrics and drug trends among American youth. Journal of Adolescent Research, 21(3),
269-298.
30
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Dissanayake, E. (2000). Antecedents of the temporal arts in early mother-infant interaction. The
origins of music, 389-410.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Harcourt Brace Jovanovich
College Publishers.
Eastman, J. T., Pettijohn, I. I., & Terry, F. (2015). Gone country: An investigation of Billboard
country songs of the year across social and economic conditions in the United
States. Psychology of Popular Media Culture, 4(2), 155.
Fiske, S. T., & Taylor, S. E. (2013). Social cognition: From brains to culture. Sage.
Gelfand, M. J., Raver, J. L., Nishii, L., Leslie, L. M., Lun, J., Lim, B. C., ... & Aycan, Z. (2011).
Differences between tight and loose cultures: A 33-nation study. Science, 332(6033),
1100-1104.
Gibson, E., Futrell, R., Piantadosi, S. P., Dautriche, I., Mahowald, K., Bergen, L., & Levy, R.
(2019). How Efficiency Shapes Human Language. Trends in Cognitive Sciences, 23(5),
389–407.
Grossmann, I., & Varnum, M. E. (2015). Social structure, infectious diseases, disasters,
secularism, and cultural change in America. Psychological Science, 26(3), 311-324.
Hayakawa, S. I. Popular songs vs. the facts of life. In B. Rosenberg & D. M. White (Eds.), Mass
culture: The popular arts in America. Glencoe, I11.: Free Press, 1957.
Henrich, J. (2015). The secret of our success: how culture is driving human evolution,
domesticating our species, and making us smarter. Princeton University Press.
Horita, Y., & Takezawa, M. (2018). Cultural differences in strength of conformity explained
through pathogen stress: a statistical test using hierarchical Bayesian
estimation. Frontiers in Psychology, 9, 1921.
Huynh, A. C. & Grossmann, I. (in press). Rising ethnic diversity in the United States
accompanies shifts toward an individualistic culture. Social Psychological and
Personality Science.
Hyden, C., & McCandless, N. J. (1983). Men and women as portrayed in the lyrics of
contemporary music. Popular Music & Society, 9(2), 19-26.
Hyndman R, Athanasopoulos G, Bergmeir C, Caceres G, Chhay L, O'Hara-Wild M, Petropoulos
F, Razbash S, Wang E, Yasmeen F (2020). _forecast: Forecasting functions for time series
and linear models_. R package version 8.12, <URL:
http://pkg.robjhyndman.com/forecast>.
31
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Iyengar, S. S., & Kamenica, E. (2010). Choice proliferation, simplicity seeking, and asset
allocation. Journal of Public Economics, 94(7-8), 530-539.
Jackson, J. C., Gelfand, M., De, S., & Fox, A. (2019). The loosening of American culture over
200 years is associated with a creativity–order trade-off. Nature Human Behaviour, 3(3),
244-250.
Jordan, K. N., Sterling, J., Pennebaker, J. W., & Boyd, R. L. (2019). Examining long-term trends
in politics and culture through language of political leaders and cultural
institutions. Proceedings of the National Academy of Sciences, 116(9), 3476-3481.
Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003). Political conservatism as
motivated social cognition. Psychological Bulletin, 129(3), 339.
Kempe, D., Kleinberg, J., & Demers, A. (2004). Spatial Gossip and Resource Location Protocols.
Journal of the ACM, 51(6), 943–967.
Khandakar, Y., & Hyndman, R. J. (2008). Automatic time series forecasting: the forecast
package for rj stat. Soft.
Kirby, S., Cornish, H., & Smith, K. (2008). Cumulative cultural evolution in the laboratory: An
experimental approach to the origins of structure in human language. Proceedings of the
National Academy of Sciences, 105(31), 10681–10686.
Kirby, Simon, Tamariz, M., Cornish, H., & Smith, K. (2015). Compression and communication
in the cultural evolution of linguistic structure. Cognition, 141, 87–102.
Krumhansl, C. L. (2002). Music: A Link Between Cognition and Emotion. Current Directions in
Psychological Science, 11(2), 45–50.
Kruskal, W. H. (1958). Ordinal Measures of Association. Journal of the American Statistical
Association, 53(284), 814–861. JSTOR.
Lambert, B., Kontonatsios, G., Mauch, M., Kokkoris, T., Jockers, M., Ananiadou, S., & Leroi, A.
M. (2020). The pace of modern culture. Nature Human Behaviour, 1-9.
Margulis, E. H. (2014). Verbatim repetition and musical engagement. Psychomusicology: Music,
Mind, and Brain, 24(2), 157.
Margulis, E. H. (2013). Aesthetic responses to repetition in unfamiliar music. Empirical Studies
of the Arts, 31(1), 45-57.
Marshall, S. R., & Naumann, L. P. (2018). What’s your favorite music? Music preferences cue
racial identity. Journal of Research in Personality, 76, 74-91.
32
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
McAllister, P. O., & Anderson, A. (1991). Conservatism and the comprehension of implausible
text. European Journal of Social Psychology, 21(2), 147–164.
Mehr, S. A., Singh, M., Knox, D., Ketter, D. M., Pickens-Jones, D., Atwood, S., Lucas, C.,
Jacoby, N., Egner, A. A., Hopkins, E. J., Howard, R. M., Hartshorne, J. K., Jennings, M.
V., Simson, J., Bainbridge, C. M., Pinker, S., O’Donnell, T. J., Krasnow, M. M., &
Glowacki, L. (2019). Universality and diversity in human song. Science, 366(6468).
Mesoudi, A., & Whiten, A. (2008). The multiple roles of cultural transmission experiments in
understanding human cultural evolution. Philosophical Transactions of the Royal Society
B: Biological Sciences, 363(1509), 3489-3501.
Morling, B., & Lamoreaux, M. (2008). Measuring Culture Outside the Head: A Meta-Analysis
of Individualism—Collectivism in Cultural Products. Personality and Social Psychology
Review, 12(3), 199–221.
Murray, D. R., Trudeau, R., & Schaller, M. (2011). On the origins of cultural differences in
conformity: Four tests of the pathogen prevalence hypothesis. Personality and Social
Psychology Bulletin, 37(3), 318-329.
Nand, K., Masuda, T., Senzaki, S., & Ishii, K. (2014). Examining cultural drifts in artworks
through history and development: cultural comparisons between Japanese and western
landscape paintings and drawings. Frontiers in Psychology, 5, 1041.
Oishi, S., Miao, F. F., Koo, M., Kisling, J., & Ratliff, K. A. (2012). Residential mobility breeds
familiarity-seeking. Journal of Personality and Social Psychology, 102(1), 149.
Pettijohn, T. F., & Sacco Jr, D. F. (2009a). The language of lyrics: An analysis of popular
Billboard songs across conditions of social and economic threat. Journal of Language
and Social Psychology, 28(3), 297-311.
Pettijohn, T. F., & Sacco Jr, D. F. (2009b). Tough times, meaningful music, mature performers:
Popular Billboard songs and performer preferences across social and economic
conditions in the USA. Psychology of Music, 37(2), 155-179.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion.
In Communication and Persuasion (pp. 1-24). Springer New York.
Piantadosi, S. T., Tily, H., & Gibson, E. (2011). Word lengths are optimized for efficient
communication. Proceedings of the National Academy of Sciences, 108(9), 3526–3529.
Rubin, D. C. (1997). Memory in oral traditions: The cognitive psychology of epic, ballads, and
counting-out rhymes. Oxford: Oxford University Press.
Reisman, D. (1957), "Listening to Popular Music", pp. 408-417, in B. Rosenberg and D.M.
White (eds.) Mass Culture, Glencoe: Free Press
33
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Rentfrow, P. J., & Gosling, S. D. (2003). The do re mi's of everyday life: the structure and
personality correlates of music preferences. Journal of Personality and Social
Psychology, 84(6), 1236.
Rentfrow, P. J., & Gosling, S. D. (2006). Message in a ballad: The role of music preferences in
interpersonal perception. Psychological Science, 17(3), 236-242.
Rentfrow, P. J., & Gosling, S. D. (2007). The content and validity of music-genre stereotypes
among college students. Psychology of Music, 35(2), 306-326.
Rycroft, R. W. (2006). Time and technological innovation: Implications for public
policy. Technology in Society, 28(3), 281-301.
Santos, H. C., Varnum, M. E., & Grossmann, I. (2017). Global increases in
individualism. Psychological Science, 28(9), 1228-1239.
Schellenberg, E. G., & von Scheve, C. (2012). Emotional cues in American popular music: Five
decades of the Top 40. Psychology of Aesthetics, Creativity, and the Arts, 6(3), 196.
Schoonvelde, M., Brosius, A., Schumacher, G., & Bakker, B. N. (2019). Liberals lecture,
conservatives communicate: Analyzing complexity and ideology in 381,609 political
speeches. PLoS ONE, 14(2).
Sng, O., Neuberg, S. L., Varnum, M. E. W., & Kenrick, D. T. (2018). The behavioral ecology of
cultural psychological variation. Psychological Review.
Steegen, S., Tuerlinckx, F., Gelman, A., & Vanpaemel, W. (2016). Increasing Transparency
Through a Multiverse Analysis. Perspectives on Psychological Science, 11(5), 702–712.
Stephens, N. M., Markus, H. R., & Townsend, S. S. (2007). Choice as an act of meaning: the
case of social class. Journal of Personality and Social Psychology, 93(5), 814-830.
Suedfeld, P. (2010). The Cognitive Processing of Politics and Politicians: Archival Studies of
Conceptual and Integrative Complexity. Journal of Personality, 78(6), 1669–1702.
Tamariz, M., & Kirby, S. (2015). Culture: copying, compression, and conventionality. Cognitive
Science, 39(1), 171-183.
Tiokhin, L., & Hruschka, D. (2017). No evidence that an Ebola outbreak influenced voting
preferences in the 2014 elections after controlling for time-series autocorrelation: A
Commentary on Beall, Hofer, and Schaller (2016). Psychological Science, 28(9), 1358-
1360.
34
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY
Tooby, J., & Cosmides, L. (2020). Natural Selection and the Nature of Communication. In K.
Floyd & R. Weber (Eds.), The Handbook of Communication Science and Biology.
Routledge.
Tybur, J. M., Inbar, Y., Aarøe, L., Barclay, P., Barlow, F. K., De Barra, M., ... & Consedine, N. S.
(2016). Parasite stress and pathogen avoidance relate to distinct dimensions of political
ideology across 30 nations. Proceedings of the National Academy of Sciences, 113(44),
12408-12413.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and
biases. Science, 185(4157), 1124-1131.
Varnum, M. E. W., & Grossmann, I. (2016). Pathogen prevalence is associated with cultural
changes in gender equality. Nature Human Behaviour, 1(1), 003.
Varnum, M. E. W., & Grossmann, I. (2017). Cultural change: The how and the why. Perspectives
on Psychological Science, 12(6), 956-972.
Varnum, M. E. W., & Grossmann, I. (2019). The wealth -> life history -> innovation
account of the industrial revolution is largely inconsistent with empirical time series data.
Behavioral and Brain Sciences, 42, e212.
Wilson, G. D., Ausman, J., & Mathews, T. R. (1973). Conservatism and art preferences.
Journal of Personality and Social Psychology, 25(2), 286–288.
Youngblood, M. (2019). Cultural transmission modes of music sampling traditions remain stable
despite delocalization in the digital age. PloS ONE, 14(2).
Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social
Psychology, 9(2), 1-27.
Zajonc, R. B. (2001). Mere exposure: A gateway to the subliminal. Current Directions in
Psychological Science, 10(6), 224-228.
35
RUNNING HEAD: CHANGES IN LYRICAL COMPLEXITY 36
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Here we investigate the evolutionary dynamics of several kinds of modern cultural artefacts—pop music, novels, the clinical literature and cars—as well as a collection of organic populations. In contrast to the general belief that modern culture evolves very quickly, we show that rates of modern cultural evolution are comparable to those of many animal populations. Using time-series methods, we show that much of modern culture is shaped by either stabilizing or directional forces or both and that these forces partly regulate the rates at which different traits evolve. We suggest that these forces are probably cultural selection and that the evolution of many artefact traits can be explained by a shifting-optimum model of cultural selection that, in turn, rests on known psychological biases in aesthetic appreciation. In sum, our results demonstrate the deep unity of the processes and patterns of cultural and organic evolution. Modern culture appears to change at a rapid rate compared with biological evolution. Lambert et al. test this and find that the pace of modern cultural evolution is surprisingly slow—slower than, for example, changes seen in Darwin’s finches.
Article
Full-text available
The majority of research on music aesthetics treats music and lyrics as discrete entities, despite the artistic imperative that they should relate to one another in some way. This research computer analyzed both the music and lyrics of the songs to have reached the weekly UK top five singles chart from January 1999 to December 2013 ( N = 1,414). The findings indicate that the typicality of a given set of lyrics relative to the corpus as a whole was associated with their popularity; that there were numerous associations between each of six mood scores assigned to the music and various aspects of the lyrics (e.g., passionate music was associated with lyrics addressing hardship and less concern with precise numerical terms); and that the relative contribution of the lyrics and music to overall popularity varied according to the means by which these were operationalized so that, for instance, music and lyrics contributed equally to explaining peak chart position, whereas music outperformed lyrics in explaining the number of weeks spent on the top five. Pop music and its lyrics are related to one another, and the relationship can be explained to some extent via existing concepts in the aesthetics literature.
Article
Full-text available
Cultural evolution of emotional expression in 50 years of song lyrics - Volume 1 - Charlotte O. Brand, Alberto Acerbi, Alex Mesoudi
Article
Full-text available
Music sampling is a common practice among hip-hop and electronic producers that has played a critical role in the development of particular subgenres. Artists preferentially sample drum breaks, and previous studies have suggested that these may be culturally transmitted. With the advent of digital sampling technologies and social media the modes of cultural transmission may have shifted, and music communities may have become decoupled from geography. The aim of the current study was to determine whether drum breaks are culturally transmitted through musical collaboration networks, and to identify the factors driving the evolution of these networks. Using network-based diffusion analysis we found strong evidence for the cultural transmission of drum breaks via collaboration between artists, and identified several demographic variables that bias transmission. Additionally, using network evolution methods we found evidence that the structure of the collaboration network is no longer biased by geographic proximity after the year 2000, and that gender disparity has relaxed over the same period. Despite the delocalization of communities by the internet, collaboration remains a key transmission mode of music sampling traditions. The results of this study provide valuable insight into how demographic biases shape cultural transmission in complex networks, and how the evolution of these networks has shifted in the digital age.
Article
Full-text available
Baumard proposes a model to explain the dramatic rise in innovation that occurred during the Industrial Revolution, whereby rising living standards led to slower life history strategies, which in turn fostered innovation. We test his model explicitly using time series data, finding limited support for these proposed linkages. Instead, we find evidence that rising living standards appear to have a bidirectional relationship with increasing innovation.
Article
Full-text available
Individualism appears to have increased over the past several decades, yet most research documenting this shift has been limited to the study of a handful of highly-developed countries. Is the world becoming more individualistic as a whole? If so, why? To address these questions, here we examine 51 years of data on individualistic practices and values across 77 countries. Our findings suggest that individualism is indeed rising in most of societies we tested. Despite dramatic shifts towards greater individualism around the world, cultural differences remain sizeable. Moreover, cultural differences are primarily linked to changes in socioeconomic development, and to a lesser extent to shifts in pathogen prevalence, disaster frequency, and climatic stress.
Article
Full-text available
More than half a century of cross-cultural research has demonstrated group-level differences in psychological and behavioral phenomena, from values to attention to neural responses. However, cultures are not static, with several specific changes documented for cultural products, practices, and values. How and why do societies change? Here we juxtapose theory and insights from cultural evolution and social ecology. Evolutionary approaches enable an understanding of the how of cultural change, suggesting transmission mechanisms by which the contents of culture may change. Ecological approaches provide insights into the why of cultural change: They identify specific environmental pressures, which evoke shifts in psychology and thereby enable greater precision in predictions of specific cultural changes based on changes in ecological conditions. Complementary insights from the ecological and cultural evolutionary approaches can jointly clarify the process by which cultures change. We end by discussing the relevance of cultural change research for the contemporary societal shifts, and by highlighting several critical challenges and future directions for the emerging field of cross-temporal research on culture and psychology.
Article
Past research has shown that music preferences can provide meaningful personality and identity clues. However, no research has examined whether music preferences can convey meaningful information regarding racial identity. Across two studies, we find racial differences in music preferences and demonstrate that Black participants with lower racial centrality are more likely to prefer music associated with White Americans. In Study 1, we find that strangers can make somewhat accurate racial judgments based on music preferences alone. In Study 2, we demonstrate that people are aware of how their music preferences reflect their race. The results suggest that people have strong racial associations with certain music genres and thus may use music preferences to communicate about their racial identities.