INTRODUCTION

SWING rhythms are often celebrated as a hallmark of jazz (Benadon, 2006; Butterfield, 2011; Friberg & Sundström, 1997), yet their significant influence on the sound of country music has received less scholarly attention, with most country music scholarship focusing on broader stylistic or cultural aspects rather than rhythmic features (Allen, 1994; Malone, 1968; Manuel, 2008; Neal, 2018). [1] Indeed, a casual listen to the discographies of notable artists like Johnny Cash, Hank Williams, and Conway Twitty reveals a pervasive presence of swing in their musical output. That said, the ‘country sound’ has changed since the genre’s early years—as discussed, for example, in the thorough overview of stylistic and cultural changes in country music by Neal (2018). Because country music has gone through such dramatic stylistic shifts over the years, and because the history of swing (a distinctive feature of rhythm and meter in country) has never been directly analyzed, this study aims to specifically examine the use and incidence of swing in the broader context of country music’s over 75-year history.

Country music’s evolution spans distinct eras that illuminate the shifting role of swing rhythms within the genre. Early styles (pre-1950s) drew heavily from folk and blues, featuring raw acoustic instrumentation and often swung rhythms, as heard in the recordings of Jimmie Rodgers and the Carter Family. The 1950s introduced rockabilly influences with upbeat, energetic feels (e.g., Elvis Presley, Carl Perkins), while the 1980s–1990s emphasized larger productions with polished arrangements and narrative storytelling (e.g., George Strait, Reba McEntire, George Jones). In recent decades, pop and hip-hop crossovers (e.g., Beyoncé, Lil Nas X, Post Malone) have emerged alongside neo-traditional revivals (e.g., Jason Isbell, Chris Stapleton). Swing rhythms, characterized by uneven subdivision of beats, often approximating a long-short pattern where the first note is roughly twice as long as the second, contrast with straight (even) rhythms. For example, straight eighth notes divide a beat evenly, whereas swung eighths approximate rhythms ranging from a triplet with the first two eighth notes tied together (and followed by a third eighth) to a dotted eighth followed by a sixteenth, with ratios varying by tempo and performer.

The incorporation of swing feel in country music has roots that stretch back to the early 20th century, reflecting a fusion of diverse cultural and musical influences ranging from African American jazz and blues to traditional folk and European dance music (Allen, 1994; Malone, 1968; Manuel, 2008). Western Swing, a subgenre of country music that emerged prominently during the 1930s and 1940s, epitomizes this hybrid sound and arguably brought the swung rhythmic feel into the greater country music genre (Neal, 2018). Swing is a prominent feature in numerous popular music genres beyond country music. Understanding its use in recorded music can hopefully provide valuable insights into broader discussions on musical notation in aspects of popular music, potentially suggesting a need to re-tool our traditional (classically-based) musical language.

The earliest historically informed analyses of swing in country music come from Jean Boyd, who likens the “Western Swing” style found in Texas and Louisiana to early jazz. She notes a “steady quadruple meter” accompanied by “unevenly subdivided beats” played on the ride cymbal (Boyd, 1988, p. 143). Boyd’s analysis establishes an important connection between Western Swing and early jazz, serving as a foundational touchpoint for the exploration of swing feel in country music. Similarly, Neal (2018) highlights the distinctive characteristics of Western Swing, emphasizing its incorporation of swing rhythms and jazz harmonies, vocabulary drawn from popular music.

While substantial musicological literature exists on the history of country music and its various subgenres, there remains a notable absence of scholarship specifically focused on analytical approaches to the country genre as a whole, let alone work that quantifies and analyzes swing and rhythmic feel within the context of country music. Rock and pop music styles often dominate popular music-focused musicological and theory research, with numerous quantitative studies (e.g., de Clercq, 2023a; Gauvin, 2015; North & Krause, 2024; Schellenberg & von Scheve, 2012; Temperley & de Clercq, 2013). Although there is an immense amount of research into ideal swing ratios and tempo in relation to swing rhythms, most of the existing analytical work is focused on jazz music (e.g., Benadon, 2006; Butterfield, 2011; Friberg & Sundström, 1997), whose use of swing rhythms is markedly more persistent throughout history. This extant scholarship suggests that there is a negative correlation between swing ratios (the duration of the on-beat note relative to the off-beat note, or “beat-upbeat” ratios) and tempo, where higher ratios (“harder” swing feel) exist most commonly at the lowest recorded tempo rates, around 75 BPM, whereas lower ratios (“lighter” or closer to even subdivisions) have an upper threshold of 320 BPM (Friberg & Sundström, 1997).

To better understand the use of swing rhythms in country music, it may help to understand broader trends in tempo across popular music. Existing corpus research reveals a consistent decline in tempo over the years across various genres, raising questions about how this shift may influence the presence of swing elements in country music. For instance, Schellenberg and von Scheve (2012) observed that the average tempo of songs decreased significantly during recent history, dropping from about 116 BPM during the period from 1965 to 1969 to about 100 BPM between 2005 and 2009.

From the existing popular music scholarship, inconsistencies within the traditional music theory system have emerged regarding meter, rhythm, and notation. The underlying complexity of this study lies in the question of whether or not a given rhythm is swung or in a compound meter; tempo plays an important role in this determination, as does surrounding musical content and genre-specific characteristics. Danielsen’s (2018) study of neo soul groove examines the local issue of meter classification when a groove has simultaneous pulses, while de Clercq (2016) highlights the global complexity of classifying meter in popular music, noting that typical approaches often fall short of the clarity found in classical theory due to the varied ways that rhythmic “feel” (e.g., half-time or double-time feel) is interpreted by listeners. In more recent literature, de Clercq (2023a) and Geary (2022) take varying approaches to the issue of the anecdotal “simultaneous tempi” felt in popular music, both of which are important to consider for any study examining rhythm and meter in a popular music genre.

While early country songwriting tends to demonstrate a rich incorporation of swing elements (as discussed above), a casual listen of modern country radio reveals a much more infrequent occurrence of swing. This apparent decline of these rhythmic influences ultimately raises critical questions about the genre’s trajectory. The correlation between tempo and swing feel, as explored in existing jazz scholarship, suggests that specific tempo ranges may be crucial for the presence of swing elements. Furthermore, the declining trends in average tempo observed in popular music imply a broader shift in rhythmic practices that could affect swing’s role in country music. This study hypothesizes that the integration of swing elements in country music has changed over the last eight decades and that distinct tempo thresholds will also characterize their use, aligning with trends noted in jazz. By analyzing these aspects, this research aims to provide a deeper understanding of the interplay between swing and country music, thereby addressing some of the gaps in current theory scholarship.

Based on this background research, I posit two distinct hypotheses:

1. With regard to the presence of swing elements in recorded country music:

• Null Hypothesis (H0): No significant changes have occurred in the use of swing elements over the course of recorded country music.

• Alternative Hypothesis (H1): Significant changes have occurred in the use of swing elements during this period.

2. Distinct upper and lower tempo thresholds will delineate the use of swing and swung subdivisions within country music recordings, similar to swing thresholds found in jazz.

Beyond these a priori hypotheses, the paper also engages in exploratory analyses to probe patterns not easily captured by simple linear models. For example, locally weighted regressions (LOESS) are used to examine potential non-linear relationships between tempo and swing usage across decades. These exploratory investigations are not formal hypothesis tests but serve to highlight trends that may merit further theoretical or historical explanation.

METHODS

Corpus Selection

The research was conducted on a dataset sampled in May of 2024. This collection includes 10 commercially recorded country songs per year from 1949 to 2023, for a total of 750 songs. [2] These songs were selected using exemplar sampling techniques (de Clercq, 2023b) from Billboard magazine’s year-end country charts, selecting the top ten songs on the chart and replacing any previously top 10-charting song with the next song further down the chart. This method was chosen for a number of reasons: 1) an uncountable number of songs are written in a given time span, making random sampling (or related methods) impossible to accurately achieve; 2) chart lengths, types, and qualifications varied greatly as Billboard evolved, limiting the length of early charts (from 1940 to the early 1960s) as compared to present-day charts, making 10 songs both the maximum number of songs to consistently select each year and a reasonably sized dataset to analyze considering the overall scale of the project; and 3) crowd-sourced compilations of early Billboard chart data are extremely inconsistent due to the above constraints and lack of archival magazines with year-end data. Early Billboard data was compiled from Joel Whitburn’s Top Country Singles, 1944-1993 (Whitburn, 1994) and The Billboard book of Top 40 country hits (Whitburn, 2006), while data post-2006 was collected from Billboard’s online “Hot Country Songs” year-end charts (Billboard, n.d.). Watson’s research (2019) on gender representation in Billboard’s country charts offers peripheral insights into trends within Billboard data over two decades, analyzing the impact of changing classification systems on the charts, ultimately representing some of the closest research to this paper in terms of quantitative country music scholarship.

Data Encoding

Encoding the musical qualities of each song into an objective numerical format proved to be quite difficult to separate from personal musical experience and biases. In order to maintain the objectivity of the study, songs were analyzed solely from audio recordings without reference to available scores. But as previous research has shown, there are potential problems with this approach. The “tactus” of a song, for instance, could possibly be felt by a listener at any of three possible tempos due to the existence of half-time and double-time drum feels (de Clercq, 2016). Other authors have noted the ambiguity between compound meter and swing, particularly at faster tempos (Wyatt et al., 2005). Because of issues like these, it was necessary to develop a standardized system that avoided subjective assessments of tactus or meter.

To account for these issues, the method used here was to first encode whether a song’s meter was organized in groups of two or three (i.e., evenly or unevenly spaced). For example, a song with a normal drum beat in 4/4, a half-time feel in 4/4, or a normal 6/8 drum beat was classified as “2”, whereas a song with a normal drum beat in 3/4 (with or without eighth-note swing) would be classified as “3”. Analysis of each song was performed where audio files were played along with a digital metronome tool to tap and measure the rate of alternation of the downbeat (strong beat) and the backbeat (also known as weak, or upbeat) in beats per minute (BPM). For the majority of songs in the dataset, this rate was determined from a kick drum onset to the following snare drum onset. For songs without a kick and snare (a smaller portion of songs), other percussive elements were considered for the upbeat onset, such as mandolin and fiddle “chucks” as well as palm-muted guitar strums. This rate does not necessarily represent the main perceived tactus of each song (such as in a song with a half-time feel) but will be referred to as the downbeat/backbeat rate throughout the rest of this paper.

Further encoding of each song determined the rhythmic makeup of the first and second subdivisions of the downbeat/backbeat rate, identifying the prominent rhythmic feel as a duple (i.e., even or “straight,” encoded as “2”) or non-duple (i.e., compound or “swung,” encoded as “3”); if a song had no audible subdivision at a particular metric level, this level was encoded as a zero. Songs with the designation of non-duple divisions (encoded as “3”) could represent either swing feel or compound meter depending on the downbeat/backbeat rate and subdivision level; mid-tempo and up-tempo songs with compound first-level subdivisions often receive the “zero” encoding for the second subdivision layer if further subdivisions proved meaningless to the rhythmic makeup of the song. The first metric level required the following: 1) all songs must be classified as one of the two rhythmic subdivisions; and 2) the presence of a non-duple subdivision automatically ruled out duple classification for that metric level. If the song utilized swing (subjectively determined) at any subdivision layer, the rate (tempo) of that subdivision layer was also recorded. This “Swung Layer Rate” can be used to quickly determine the presence of songs with non-duple subdivisions, as seen below, as well as look at tempo trends within swing usage. For the sake of illustration, Table 1 presents some raw data from the dataset and interprets the data using conventional musical terminology. Interestingly, the full dataset includes no songs in 3/4 that utilize only duple subdivisions.

To evaluate the hypotheses and explore related patterns, the analyses were divided into two categories. A priori tests were conducted to address the stated research questions. These included chi-squared tests to examine associations between year of release and the use of duple or non-duple subdivisions (testing H1), linear regressions to evaluate the relationship between year and downbeat/backbeat tempo (testing H2 indirectly via tempo trends), and descriptive statistics comparing tempo distributions across subdivision categories (directly testing tempo-subdivision relationships in H2). In addition to these planned analyses, exploratory tests were also carried out to identify trends that might not be well captured by simple linear models. For example, locally weighted regressions (LOESS) were applied to visualize potential non-linear changes in tempo and regression contrasts were used to compare duple-only and swung subdivision classes. These exploratory analyses were not part of the original hypotheses but provide additional context for interpreting the primary results.

RESULTS

All data analyses were performed with Python and standard statistical modeling libraries. Of the 750 analyzed songs from 283 different artists during the timespan of 1949–2023, more than half feature purely duple subdivisions (2, 2; e.g., straight 4/4 meter), as show in Table 2. A second, smaller set of songs utilizes compound subdivisions at the first metric level with negligible second-level subdivisions (3, 0; e.g., 4/4 with swung 8ths), while the remaining categories collectively account for less than ten percent of the sample.

Swing Usage Over Time

In order to examine possible relationships between the year of a song’s release and the presence (as well as subdivision levels) of non-duple rhythms, chi-squared tests were employed. Chi-squared tests are particularly useful because they assess whether the distribution of a categorical variable—in this case, the duple/non-duple or “straight” vs. “swung” attributes of the dataset—occurs by chance or reflects an uneven distribution, which here is assessed across the time periods of the study. Under the null hypothesis, we would expect a roughly even distribution over time, but Figure 1 shows an uneven distribution of the subdivision characteristics. The chi-squared results between “Year” and “Duple/Non-Duple Subdivisions” (for the entire sample, regardless of subdivision level) were statistically significant, X² (74, N = 750) = 389.42, p < .001, V = .72, which suggests that the occurrence of songs with non-duple subdivisions or a “swung” feel is not constant over time.

To further investigate possible trends suggested by the uneven distribution found in Figure 1, the swing usage hypothesis can be tested by filtering the data for songs with non-duple subdivisions and plotting the utilization percentage in a histogram, as shown in Figure 2. As the red regression line in Figure 2 indicates, there is a significant decrease in compound subdivision or swing utilization over the course of the study period. Following an initial peak, the regression line indicates a decline in swing usage, with approximately 77% percent of the variation in swing percentage explained by the year (r² = .77, p < .001), despite notable fluctuations from year to year, specifically in the later decades. Several years show extremely low swing usage, including eight years with zero recorded non-duple rhythms (or 10 songs/year with only duple/straight rhythms) in the 1990s and 2000s.

As shown in Figure 3, two additional chi-squared tests were used to determine the association between year and rhythmic subdivisions at the two encoded metric levels (1st and 2nd, respectively). The results for “Year” and “1st Level Subdivision,” X² (74, N = 750) = 421.92, p < .001, V = .75, provide evidence of a significant relationship between the year a song was released and the type of subdivisions that are present in the song, similar to the overall “Year” and “Duple/Non-Duple Subdivision” set (Figure 1), where there is an even stronger temporal relationship between the year and subdivision patterns (as evidenced by the higher chi-squared value). The second level subdivision calculation required slight modification, where the “0” and “2” values for negligible and duple subdivisions were combined, since they both essentially represent no swing at the second subdivision level. This results, X² (74, N = 750) = 65.22, p = 0.76, V = .30, which suggests no statistically significant relationship between year and the use of duple or non-duple subdivisions at the second metric level, implying that the null hypothesis cannot be rejected. Overall, the results shown in Figure 3 suggest that while swing usage and the primary subdivisions (2, 3) at the first metric level are closely linked to temporal trends, the subdivisions at the second metric level are not clearly tied to the release year of a song.

To further examine possible trends in the uneven distributions of Figure 3, linear regressions were performed on the utilization of non-duple rhythms at the first and second level subdivisions. In the lefthand panel of Figure 4, the first level non-duple subdivisions (e.g., if the downbeat/backbeat rate were quarter notes, these songs would utilize swung eighths or a 6/8 meter) become less common over the period (r² = .75, p < .001), with similar proportions to the overall non-duple subdivisions category (Figure 2). In the righthand panel of Figure 4, the second level non-duple subdivisions (e.g., swung 16th notes) are non-existent until the late 1960s, where they then account for 20% or less of the sample in any given year.

To further explore this discrepancy, a final chi-squared test was used to compare the “1st Level Subdivision” and “2nd Level Subdivision” categories (see Table 2), which demonstrated a significant interaction between the first and second levels of rhythmic subdivision in the dataset: X² (2, N = 750) = 611.12, p < .001, V = .90. The strong Cramer’s V of .90 suggests that rhythmic subdivisions at the two metric levels are tightly interconnected, which may explain why the previous chi-squared tests of non-duple rhythms against time (Figure 1) and first level subdivisions against time (Figure 3, left) were significant when the second level subdivisions (Figure 3, right) had such a weak relationship. Ultimately, this indicates that songs following a duple or compound rhythm at the first metric level often display specific corresponding structured differences at the second metric level. Most commonly, these present as songs analyzed as entirely duple/straight, compound or swung at the first level, straight at the first metric level and swung at the second subdivision level, or in a compound meter and swung at the second subdivision level (as seen in Table 2).

Tempo Trends

To examine the second hypothesis regarding tempo distribution and possible changes in the dataset over time, it is useful to plot the analyzed tempo rates on a scatterplot with a linear regression to approximate general trends. When using the downbeat/backbeat rate for all songs by year, the linear regression shown in the lefthand panel of Figure 5 resulted in a negative trend, or a gradual decrease in the downbeat/backbeat rate over the 75-year period with a moderate-to-large correlation (r² = .19) between the variables and statistically significant p values (p < .001). Since tempo changes may not strictly follow a linear pattern, a locally weighted polynomial (LOESS) regression is shown in the right panel of Figure 5. The LOESS regression highlights a slight increase in tempo in the late 1980s and early 1990s that is somewhat obscured by the linear regression.

To examine a possible correlation between song tempo and different subdivision usage over time, linear regressions were performed on the backbeat rate of songs with non-duple subdivisions at either (or both) metric layer(s), as shown in Figure 6 (left). This resulted in moderate-to-large effect between the downbeat/backbeat rate and year (r² = .16), with a statistically significant p value (p < .001). When analyzing the remaining data, the songs with entirely duple subdivisions (Figure 6, right) resulted in a slightly smaller effect (9%) between the downbeat/backbeat rate and year, again with a statistically significant p value (r² = .09, p < .001). The two regressions of the downbeat/backbeat rate in Figure 5 and Figure 6 (left) display declining trends similar to Schellenberg and von Scheve (2012), but with moderate-to-large r² values in combination with significant p values, whereas the r² values from their 2012 dataset showed weaker effect in combination with significant p values, similar to the r² value in Figure 6 (right; Ellis, 2010).

Noticeably, the regression of the duple (straight) song class (Figure 6, right) had a significantly (p < .001) lower slope coefficient (β̂= –0.68, meaning a decrease of 0.68 BPM per year) than the non-duple class (β̂ = –1.04, a decrease of 1.04 BPM per year) in the lefthand panel of Figure 6. When filtering the regression further to look at 1970–2023 (where the bulk of the straight class data lies), the slope is (expectedly) less extreme (β̂ = –0.56), suggesting that the downbeat/backbeat rate of duple country songs has changed less over the course of the study than Figure 5 initially suggests, or that the regression is overwhelmingly controlled by the tempo characteristics of specific song subsets (i.e., swung first-level subdivision songs are typically faster, swung second-level subdivision songs are typically slower). Because the downbeat/backbeat rate of the duple class never slows down as drastically as other classes, this may explain why songs with a non-duple second-level subdivision (swung at the second metric layer)—those we would anecdotally expect to exist most commonly at slower tempos—never become prevalent in the dataset. Furthermore, this could explain the consistent occurrence of low-effect r² values in combination with significant p values (Figures 4, right, and 6, right).

Subdivision Interaction

To examine the relationships between swing usage, subdivision levels, and tempo over time, I conducted another analysis of the dataset stratified by subdivision pairs, as shown in Table 3. The songs with entirely duple rhythmic subdivisions have a mean downbeat/backbeat rate of 104 BPM, while the songs with a swung first level subdivision (and negligible second level subdivision) have a mean downbeat/backbeat rate of 144 BPM, indicating that on average, swung songs tend to have a relatively brisk tempo (as compared to a standard tempo of 120 BPM). Furthermore, the interquartile range of the swung first-level subdivision set extended from 110 to 177 BPM, while the entire range spanned from 72 to 304 BPM, which would align with the higher frequency of double-time rhythmic feels in the instrumentation of these songs. The 42 songs with swung second level subdivisions (the sum of the rightmost column in Table 2) had a mean downbeat/backbeat rate of 81 BPM with an IQR extending from 68 to 91 BPM, indicating a much slower tempo range for songs with swung 16th notes and/or compound meter with swung 16th notes than songs with swung first level subdivisions.

When analyzing the swung layer rate (the rate at which the non-duple subdivision exists) in Table 4, the effect of the study’s subdivision structure becomes evident. The 42-song set of swung second level subdivisions with a mean swung layer rate of 168 BPM is approximately double the downbeat/backbeat rates seen in Table 3. This doubling reflects the “swung” layer occurring twice as fast as the downbeat/backbeat, even though the downbeat/backbeat rate tends to be slower than that of songs with swung first subdivisions. In contrast, the (3, 0) subdivision pair (n = 250) aligns closely with the downbeat/backbeat rates in Table 3 (within plus or minus 1 BPM). This similarity makes sense, since the swing in these songs happens at the first subdivision level, mirroring the tempo of the downbeat/backbeat rate rather than creating an additional layer of rhythmic information. While the interquartile ranges and standard deviations for the different sets suggest considerable variation, comparison of the summary statistics for the swung layer rate reveal similar mean rates for all of the subdivision pairs except the (3, 2) pair, which consists of the compound meter songs in the sample. A closer look at the median rates for those sets shows more variability, from 133 BPM to 175 BPM, but ultimately suggests that the swung layer tends to exist generally between 130 and 180 BPM in the dataset. Furthermore, the average swung layer rate for songs with second-level subdivisions tends to be slightly faster than that of songs with swung first-level subdivisions.

When contextualized with the broader tempo findings and tests above, this examination reveals that songs with compound first-level subdivisions exhibit a significantly greater tempo, as reflected by the higher mean downbeat/backbeat rate and swung layer rate compared to the duple rhythm songs. Conversely, all songs with swung second-level subdivisions show distinctly lower downbeat/backbeat rates, indicating that when swing is used at a finer subdivision level, it tends to slow down the downbeat/backbeat tempo (or that when the downbeat/backbeat tempo decreases, swing is used at finer subdivision levels). Interestingly, the swung layer rate is slightly higher for the second-level subdivision songs than first level subdivisions. Moreover, the wide range of downbeat/backbeat rates within the swung/compound first-level subdivision songs (from 72 to 304 BPM) compared to the narrower range in the swung second-level subdivision songs (from 41 to 120 BPM) suggests that first-level swing allows for greater tempo flexibility. When considering these ranges, the rhythmic “octaves” of the swung first-level subdivision songs is about 3:1, whereas the swung second-level subdivision songs is about 4:1—different, but not quite as large as the raw data suggests.

In summary, the analysis demonstrates distinct patterns in tempo and swing usage across different rhythmic subdivisions. Songs with duple rhythms maintain moderate tempos, while those with swung/compound first-level subdivisions tend to adopt faster tempos, and those with swung second-level subdivisions generally exhibit slower tempos when considering downbeat/backbeat rates. The variations in downbeat/backbeat rates and swung layer rates underscore the influence of swing and subdivision levels on the rhythmic and temporal characteristics of songs and ultimately demonstrates the tempo range where swung rhythms are most heavily used in country music in the years studied.

GENERAL DISCUSSION

In examining the relationship between swing usage and tempo thresholds within top-10 charting country music recordings, this study initially hypothesized that distinct upper and lower tempo limits would delineate swing application. However, the findings suggest that such straightforward thresholds do not universally apply to all swing characteristics in country music. Instead, the data reveal a more nuanced picture. Specifically, swing at the first subdivision level—often represented by eighth-note swing—tends to occur most frequently within a certain tempo range, generally between 110 and 176 BPM. This indicates that while there are broad tempo trends associated with swing characteristics, the exact tempo range where swing is applied varies. For instance, swing feels are more prevalent in the upper end of this range, suggesting a correlation between faster tempos and the likelihood of incorporating swing at the first subdivision level. However, this relationship is not absolute, as the study also uncovered examples of swing elements occurring outside this range.

Fifteen of the downbeat/backbeat rates recorded in the dataset are greater than 240 BPM, which is generally considered the highest tempo at which humans can usefully perceive a beat (London, 2004). While these recorded rates do not necessarily represent the tactus, they do highlight the challenges in accurately categorizing and analyzing tempos that exceed perceptual limits. This discrepancy points to the need for a standardized system for tempo categorization that can better account for different rhythmic feels and provide a more accurate framework for popular musical analysis. Such a system would help mitigate the challenges posed by extremely high (or low) tempos, ensuring that analyses remain relevant and interpretable. By recognizing the interaction between various rhythmic feels and tempo, both musicians and researchers alike can contribute to a more concise musical vocabulary. This, in turn, could enhance the consistency and comparability of studies across different musical genres and time periods, leading to a better understanding of how tempo and rhythm interact in popular music.

The first hypothesis, proposing a significant decrease in the presence of swing elements in country music over the nearly eight-decade span, is supported by the data. The analysis reveals a consistent negative trend in the use of swing feel, particularly from the 1980s onwards. The prevalence of swung first-level subdivisions from 1949 to 1990, as shown in Figure 6, reflects a period when swing rhythms were a defining feature of country music, often accompanied by high-energy, double-time rhythmic feels. This decline is evidenced by a noticeable reduction in songs featuring compound or swung rhythms, alongside comparatively large fluctuations in downbeat/backbeat rates. Such a trend aligns with broader changes in musical style and production practices both within country music and in other musical genres. The decrease in swing elements could reflect shifts in audience preferences or changes in musical trends that favor more straightforward rhythmic structures. Another suggested factor in the decline of swing (and the rise of 16th note swing) is attributed to the rise of digital audio workstations (DAWs) and drum machines, which may have originally limited swing capabilities before adding newer capabilities like 16th note swing that were extremely rare in the early years of the dataset.

Despite its comprehensive approach, the study has several notable limitations. While the dataset covers a substantial period, it is not immune to selection biases inherent in chart-based sampling methods. The reliance on Billboard’s year-end country charts, though practical for ensuring a long-term source of popular songs, may inadvertently introduce biases related to the chart lengths and methods over time. This sampling method might not fully capture the diversity of rhythmic practices across all country music, particularly for less commercially successful songs that did not make the charts but might exhibit swung rhythmic characteristics. To address this, larger samples could be analyzed using weighted statistical methods to overcome chart length limits, assuming sufficient resources are available to handle the increased volume of recordings. Additionally, the lack of standardized notational methods for capturing tempo and rhythmic feels poses significant challenges when comparing results to other quantitative music analyses. Rhythmic structures in music, especially those involving nuanced swing and compound meters, are often subject to interpretive variability, e.g., a slow compound rhythm sped up could transform into a swung rhythm and vice versa. The study’s encoding process, which attempts to ignore bar length and music notation, is quite abstract from the vocabulary of “8th note swing” or “6/8 with swung 16th notes,” requiring subjective translation to determine the results of a 180 BPM backbeat rate with a compound first level subdivision, for example. These limitations underscore the inherent challenges in analyzing rhythmic structures and point to the need for more refined notation and transcription systems for popular music in the future.

To address the questions posed by the decline of swing and better understand the observed trends, future research should focus on analyzing larger sample sizes and more granular data from the latter half of the dataset. Specifically, the years 1980 to 2010 spark additional interest due to both the lack of songs with compound or swung rhythms present and the slight increase and then decrease in downbeat/backbeat rates. Additionally, the most recent decade in the dataset shows a slight increase in the number of country songs with swing present after a dry spell without swung rhythms. This resurgence raises questions about whether the dataset represents a genuine trend or if it is a result of sampling biases or other factors. Larger sample sizes for the latter half of the dataset may provide a better indication of whether the current model is truly representative of the country music landscape in more recent years.

In conclusion, this study attempts to provide a first quantitative analysis of swing usage and tempo trends in country music over a span of nearly eight decades. By analyzing the relationships between swing, rhythmic subdivisions, and tempo, the research offers new insights into the genre’s rhythmic evolution. The findings reveal distinct patterns in how swing and tempo interact, highlighting the complex interplay between these elements in shaping the rhythmic characteristics of country music.

Acknowledgements

This paper would not be possible without the support of Dr. Trevor de Clercq who both inspired and advised this research inquisition. Additional thanks to Dr. Richard Hoffman and Dr. Clare Eng of Belmont University who wholeheartedly support theory research in undergraduate curriculums. This article was copyedited by Jonathan Tang and layout-edited by Niels Chr. Hansen.

REFERENCES

Allen, B., ed. (1994). The Blackwell Guide to Recorded Country Music. Oxford, UK: Blackwell.

Benadon, F. (2006). Slicing the beat: Jazz Eighth-Notes as Expressive Microrhythm. Ethnomusicology, 50(1). https://doi.org/10.2307/20174424

Billboard. (n.d.) “Year-end charts: Top country songs.” Accessed September 16, 2024. https://www.billboard.com/charts/year-end/hot-country-songs/

Boyd, J. A. (1998). The jazz of the southwest: An oral history of western swing (1st Ed.). Austin, TX: University of Texas Press. https://doi.org/10.7560/708594

Butler, C., & Butler, P. (1963). Don’t let me cross over [Song]. On Don’t let me cross over [Album]. Columbia Records.

Butterfield, M. W. (2011). Why do jazz musicians swing their eighth notes? Music Theory Spectrum, 33(1), 3–26. https://doi.org/10.1525/mts.2011.33.1.3

de Clercq, T. (2016). Measuring a measure: absolute time as a factor for determining bar lengths and meter in pop/rock music. Music Theory Online, 22(3). https://doi.org/10.30535/mto.22.3.3 

de Clercq, T. (2023a). Tempo versus average rates of harmonic rhythm in popular music: a study of three corpora. Musicae Scientiae, 27(3), 521–540. https://doi.org/10.1177/10298649221091483 

de Clercq, T. (2023b). Representation in corpus studies of music: Commentary on Shea’s (2022) “A demographic sampling model and database for addressing racial, ethnic, and gender bias in popular-music empirical research.” Empirical Musicology Review, 18(2), 175–180. https://doi.org/10.18061/emr.v18i2.9726

Ellis, P. D. (2010). The essential guide to effect sizes: Statistical power, meta-analysis, and the interpretation of research results. New York, NY: Cambridge University Press. https://doi.org/10.1017/CBO9780511761676

Ford, T. E. (1960). Mule train [Song]. On Sixteen tons [Album]. Capitol Records.

Friberg, A. & Sundström, A. (1997). Preferred swing ratio in jazz as a function of tempo. TMH-QPSR, 38(4), 19–27. https://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-234423

Gauvin, H. L. (2015). “The times they were a-changing’”: A database-driven approach to the evolution of harmonic syntax in popular music from the 1960s. Empirical Musicology Review, 10(3): 215–238. https://doi.org/10.18061/emr.v10i3.4467 

Geary, D. (2022). Analyzing the beat in metrically consonant popular songs: A multifaceted approach. Music Theory Online, 28(4). https://doi.org/10.30535/mto.28.4.2

Houston, D. (1969). Baby, baby (I know you’re a lady) [Song]. On Baby, baby (I know you’re a lady) [Album]. Epic Records.

Kingsbury, P. (1998). The encyclopedia of country music: The ultimate guide to the music. New York, NY: Oxford University Press.

London, J. (2004). Hearing in Time: Psychological Aspects of Musical Meter. New York, NY: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195160819.001.0001

Malone, B. C. (1968). Country music, U.S.A.: A fifty-year history. Austin, TX: University of Texas Press.

Manuel, J. T. (2008). “The Sound of the Plain White Folk? Creating Country Music’s ‘Social Origins.’” Popular Music & Society, 31(4), 417–431. https://doi.org/10.1080/03007760802052551

Miller, R. (1965). King of the road [Song]. On The return of Roger Miller [Album]. Smash Records.

Neal, J. R. (2018). Country music: A cultural and stylistic history. New York, NY: Oxford University Press.

North, A. C., & Krause, A. E. (2024). Are the beatles different? A computerized psychological analysis of their music. Empirical Musicology Review, 18(2), 133–147. https://doi.org/10.18061/emr.v18i2.9178 

Owens, B. (1966). Think of me [Song]. On Open up your heart [Album]. Capitol Records.

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), 196–203. https://doi.org/10.1037/a0028024

Temperley, D. & de Clercq, T. (2013). Statistical analysis of harmony and melody in rock music. Journal of New Music Research, 42(3), 187–204. https://doi.org/10.1080/09298215.2013.788039 

Watson, J. (2019). Gender on the Billboard hot country songs chart, 1996-2016. Popular Music and Society, 42(5), 538–560. https://doi.org/10.1080/03007766.2018.1512070 

Whitburn, J. (1994). Joel Whitburn’s top country singles, 1944–1993. Menomonee Falls, WI: Record Research Inc.

Whitburn, J. (2006). The Billboard book of top 40 country hits. New York, NY: Billboard Books.

Wyatt, K., Schroeder, C., & Elliott, J. (2005). Ear training for the contemporary musician. Milwaukee, WI: Hal Leonard.

Zimmerman, B. (2022). Rock and a hard place [Song]. On Leave the light on [Album]. Warner Music Nashville.